Graded expression of Emx-2 in the adult newt limb and its corresponding regeneration blastema1

Article No. mb981782

J. Mol. Biol. (1998) 279, 501±511

Graded Expression of Emx-2 in the Adult Newt Limb and its Corresponding Regeneration Blastema Michel Beauchemin1, Katia Del Rio-Tsonis2, Panagiotis A. Tsonis2 Monique Tremblay1 and Pierre Savard1* 1

Neuroscience Research Unit University Hospital Research Center of QueÂbec, CHUL Pavilion, 2705 Laurier blv Sainte-Foy, Canada G1V 4G2 2 Department of Biology The University of Dayton 300 College Park, Dayton OH 45469-2320, USA

Amputation of a newt limb causes stump cells to organize the reformation of the missing structures. The phenomenon is remarkably precise in that the regeneration is perfect. During the ®rst few days following amputation, the tissue proximal to the plane of amputation gives rise to the blastema, an area of growth composed of mesenchymal cells covered by a single epithelium. The blastema possesses a morphogenetic potential characteristic of the structures that have been amputated. Looking for control genes putatively involved in regeneration, we cloned the newt version of the mouse and human Emx-2. Its expression is restricted to the skin of the regeneration territories and is graded along the proximal-distal axis of both forelimb and hindlimb, with higher levels in distal regions. The regeneration blastema also show this proximal-distal graded level of expression with distal blastemas (mid-radius and ulna) showing higher levels of expression when compared to blastemas of more proximal origin (mid-humerus). Finally, retinoic acid proximalizes both the level of Emx-2 expression and the positional memory of the blastema suggesting Emx-2 may participate in pattern formation by specifying positional information. # 1998 Academic Press Limited

*Corresponding author

Keywords: Emx; positional information; regeneration; urodele; newt

Introduction Urodela amphibians, such as the newt Notophthalmus viridescens can replace amputated portions of limb and tail by the process of epimorphic regeneration (for a review, see Wallace, 1981; Tsonis, 1996). Immediately after amputation, epidermal cells of the stump edge migrate rapidly to cover the wound and form the apical ectodermal cap. Then the underlying tissue becomes disorganized and dedifferentiates to a common mesenchymatous-like cell type. These cells acquire a high proliferation rate, which, in conjunction with the dedifferentiation process, leads to a conical-shaped accumulation of multipotent cells, called the blastema. During subsequent days, the blastema grows rapidly and progressively differentiates from its proximal region to undergo morphogenesis, leading to the perfect replacement of the missing structures. Abbreviations used: RA, retinoic acid; PCR-RACE, polymerase chain reaction and rapid ampli®cation of cDNA ends; DMSO, dimethylsulfoxide. 0022±2836/98/230501±11 $25.00/0

A few regions of the newt body, referred to as the regeneration territories including the limbs, the tail, the snout, and the dorsal ®n, are capable of epimorphic regeneration (GuyeÂnot et al., 1948; Kiortsis, 1953). Other body parts that do not express epimorphic regeneration capabilities, such as the ¯ank, are known as neutral territories. Some intrinsic properties of the cells in a regeneration territory enable their reversal of differentiation, their re-entry in cell cycle, and their participation in the formation of a blastema (Ferretti & Brockes, 1991; Lo et al., 1993; Tanaka et al., 1997; Brockes, 1997), as though they have the morphogenetic capabilities of embryonic cells present in morphogenetic ®elds. Indeed, the regeneration territories of the adult newt and some embryonic ®elds have similar morphogenetic capabilities (Muneoka & Bryant, 1984; Muneoka & Sassoon, 1992). For example, both tissues allow the regulative development of a predictable body structure. It would not be surprising to ®nd in both tissues the expression of genes particular to morphogenetic ®elds. The blastema is an autonomous system that always gives rise to the structure of its origin, # 1998 Academic Press Limited

502 suggesting that blastema cells possess a memory of the type of appendage to form (reviewed by Stocum, 1984). This was demonstrated by the transplantation of limb blastema onto X-ray irradiated tail, where the blastema regrew a limb-like appendage. As predicted, the converse experiment of transplanting a tail blastema onto an X-ray irradiated limb led to the growth of a taillike appendage. Moreover, amputation of a limb at any level along the proximal-distal axis will always produce a faithful regeneration of the missing parts only. Amputation at the level of the wrist induces the reformation of a paw, whereas amputation at the level of the elbow induces the reformation of a forearm and a paw. Therefore, the regeneration blastema also possesses a memory of its position along the appendage (reviewed by Stocum, 1984). Retinoic acid (RA) and its derivatives are the only chemicals known to affect the determination of blastema cells; among its effects, it proximalizes the proximal-distal, memory of a limb blastema. Thus, instead of forming a paw, a distal blastema treated with RA will regenerate a more proximal structure (humerus, radius, ulna) in a dose-dependent fashion (Niazi & Saxena, 1978; Maden, 1982). The action of RA depends on nuclear receptors (Brockes, 1996; Gann et al., 1996; Pecorino et al., 1996a,b), which bind to the regulatory region of target genes, to modulate their expression. The effects of RA on homeobox gene expression during development could account for the effects of RA on pattern formation (IzpisuÂa-Belmonte et al., 1991; Kessek & Gruss, 1991; Nohno et al., 1991; Kessel, 1992; Marshall et al., 1993). Homeobox-containing genes code for transcription factors that are involved in the determination of the different structures of the body (reviewed by McGinnis & Krumlauf, 1992). Many are expressed during limb development, with expression patterns suggesting their involvement in the spatial patterning of the anterior-posterior (Dolle et al., 1989; Oliver et al., 1989; IzpisuÂa-Belmonte et al., 1991, 1992; Ros et al., 1992; Peterson et al., 1994), the proximal-distal (Yokouchi et al., 1991; Peterson

Regulation of Emx-2 in Regeneration

et al., 1994), and the dorsal-ventral axes (Davies et al., 1991; Peterson et al., 1994). The homeobox genes Emx-2 and Emx-1 are vertebrate homologs of a Drosophila head gap gene ems (Simeone et al., 1992a,b). Homozygous Emx-2 mutant mice display defects in development of the dorsal telencephalon (Yoshida et al., 1997) and mice die due to failure of urogenital system development (Miyamoto et al., 1997). Emx-2 is also expressed in the epidermis of the developing mouse limbs (Simeone et al., 1992a) and its function in limb development is not known yet. On the basis of limb development and limb regeneration possibly sharing some genetic controls, we looked for the expression of homeobox containing genes in the newt regeneration territories and their corresponding blastemas. Here, we report the isolation of Emx-2. We have found that its primary structure is strongly conserved among vertebrates. Emx-2 expression was analyzed in different regions of the body and along the proximal-distal axis of the limb corresponding blastema, either treated or not treated with RA. The results are discussed in relation to the putative role of Emx-2 in the speci®cation of the proximal-distal memory of the blastema.

Results Cloning A redundant oligonucleotide probe (U-box), speci®c for the third helix region of homeodomains, was used to screen a newt tail cDNA library (Beauchemin & Savard, 1992). The p027 clone (Figure 1(A)) included 56 of the 60 amino acids of a homeodomain. An additional cDNA library screening with p027 did not reveal other cDNA of the same family. The cloning was extended in the 50 region of the homeodomain by PCR-RACE (Frohman et al., 1988; Frohman, 1990; Jain et al., 1992). Clones TA ‡ 54, TA ‡ 159, and TA ‡ 217 showed 50 -extensions of 54, 159, and 217 bp, respectively (Figure 1(A)). A Southern blot of newt genomic DNA was hybridized with p027. HindIII digested sample

Figure 1. Schematic representation of Emx-2 cDNA clones and Southern blot analysis of newt genomic DNA. (A) the cDNA clone p027 was isolated from a newt tail cDNA library. The clones TA ‡ 54, TA ‡ 159, and TA ‡ 217 were obtained by PCR-RACE ampli®cation of newt limb cDNA; they extend p027 by 54, 159, and 217 bp, respectively. Open boxes represent coding regions whereas hatched boxes represent the homeodomain. B, BamHI; P, PvuII; Sm, SmaI; Ss, SspI; St, StuI. (B) Southern blot analysis of newt genomic DNA with p027 cDNA as a probe. Lane 1, undigested DNA (10 mg); lane 2, EcoRI digested DNA (25 mg); lane 3, HindIII digested DNA (25 mg). The molecular mass markers are the fragments of l-DNA digested with HindIII.

503

Regulation of Emx-2 in Regeneration

revealed a single band whereas EcoRI digested sample revealed two bands (Figure 1(B)). The intensity of the bands in the EcoRI digest are approximately 50% the level of the HindIII band, suggesting that an EcoRI restriction site is located in the genomic DNA harboring the p027 sequences. Since there is no EcoRI restriction site in p027, we believe Emx-2 is harboring an intron. Consistent with this interpretation, a splicing site is known to be present in the homeodomain of the human and mouse versions of the same gene (Simeone et al., 1992a). We conclude that Emx-2 is a single-copy gene. Computerized data bank searches revealed that the newt Emx-2 homeodomain shares a high degree of homology with the corresponding region in the Drosophila ems (Dalton et al., 1989). Moreover, 131 of the 135 amino acids of the coding region are identical to an equivalent region in the mouse and human EMX-2 (Simeone et al., 1992a; Figure 2). We concluded these sequences are versions of a strongly conserved gene in vertebrates. The similarities between newt Emx-2 and other vertebrate homologs include a domain located upstream from the homeodomain that is a divergent version of the conserved homeopentapeptide Ile/Phe-Tyr-Pro-Trp-Met present in several vertebrate homeobox genes belonging to the HOX clusters (Scott et al., 1989), the two residues Arg-Lys immediately upstream of the homeodomain, and the three acidic residues Glu-Glu-Glu immediately downstream of the homeodomain.

Expression in the adult newt appendages Emx-2 expression in different body regions of the newt was analyzed on Northern blots (Figure 3). A single rare transcript of approximately 2.5 kb was revealed in RNA samples of tail, hindlimb, forelimb, and their corresponding blastema, whereas there was no signal in the ¯ank. Emx-2 was also found to be expressed in the adult kidney (as for many other homeobox containing genes), whereas it was not in liver and spleen (data not shown). Several normalizing genes were used for the complementary information they provide on RNA integrity and origin (muscle or skin). A Xenopus cytokeratin probe revealed three messenger RNAs (3.5, 3.2, and 2.8 kb) that are speci®cally expressed in the epidermis (Mathisen & Miller, 1987; Ferretti et al., 1991). The intensity of the signal (Figure 3) demonstrated that the proportion of RNA coming from the epidermis is quite similar in samples of tail, hindlimb, ¯ank, and forelimb, whereas it is approximately two times higher in blastema samples. Xenopus actin probe (Mohun et al., 1984) revealed a 1.4 kb and a 1.8 kb transcript (low stringency hybridization) corresponding to muscular (a-sarcomeric) and cytoplasmic (b-cytoplasmic) actin, respectively. Analysis of the intensity of the signal (Figure 3) showed that RNA samples from different body structures, like the limb and tail, contain different ratios of muscular-to-cytoplasmic actin whereas blastema tissue only expresses cytoplasmic actin. Therefore, actin is a probe of choice

Figure 2. Nucleotide sequence of the newt Emx- cDNA and its conceptual translation product. The homeodomain is boxed. In the coding region, 131 amino acids are identical to the human Emx-2, and four amino acids are different; these are labeled with stars. Three conserved peptide motifs (see the test) are underlined.

504

Figure 3. Body distribution of Emx-2 in the adult newt. A Northern blot loaded with 5 mg of poly(A)‡RNA (absorbance at 260 nm) of various samples was serially hybridized with Emx-2, cytokeratin, and actin probes. T, tail; BT, tail blastema; HL, hindlimb; Fk, ¯ank, FL, forelimb, BL, forelimb blastema. The numbers at right of each panel indicate the apparent length of the transcripts.

to normalize RNA samples prepared from different harvests of blastemas. In situ analysis in regenerating limb revealed Emx-2 was mainly expressed in the newt limb skin and the wound epithelium of the regenerate (Figure 4). The restricted expression of Emx-2 in the skin fraction was con®rmed on Northern blot (Figure 5). Newt hindlimbs, forelimbs, and their corresponding blastemas were mechanically dissected to separate skin and muscle (intact limbs), or mesenchyme and epidermis (regeneration blastemas); the tissues were frozen and kept in liquid nitrogen before poly(A)‡RNA extraction. The skin fraction of adult hindlimb contained epidermis, dermis, and connective cells, whereas the muscle fraction contained almost exclusively muscular cells. On the other hand, the blastema is composed of a cone of mesenchymal cells and a layer of covering epidermis; the separation of these tissues from each other is dif®cult and can only result in enriched fractions with more or less contamination of the other tissue fraction. A Northern blot, loaded with samples of poly(A)‡RNA of the different tissue fractions, was serially hybridized with cytokeratin, dlx-1 (NvHBox-4), Hox C6 (NvHBox-1), actin, and Emx-2 probes (Figure 5). As expected, the cytokeratin signal was enriched in the epidermis fraction, whereas it was almost not detected in the mesenchymal fraction of the regeneration blastema; the low intensity signal may represent the level of epidermal contamination of the mesenchymal sample or the mesenchymal expression of a blastema speci®c cytokeratin (Ferretti et al., 1991). dlx-1 is a distal less like (dll-like) homeobox-con-

Regulation of Emx-2 in Regeneration

taining gene that is expressed ubiquitously in the skin fraction of the adult newt limbs (Beauchemin & Savard, 1992). Its pattern of expression in different tissue fractions of the limb and its corresponding blastema is identical to cytokeratin (Figure 5). The newt Hox C6 (Savard et al., 1988; Savard & Tremblay, 1995) expressed a 1.8 kb transcript in the muscle and skin fractions of the adult newt limb and in the epidermis and mesenchymal fractions of blastemas (Figure 5). The signals revealed with the actin probe show the enrichment of the muscular actin transcript in the muscle fraction, and the cytoplasmic one in the skin fraction of the limb (Figure 5). In the blastema, the cells are not yet differentiated, and the level of cytoplasmic actin is representative of all the cells in the blastema, including its epidermis. Finally, the expression of Emx-2 was similar in limb skin of ventral or dorsal origin when using the cytokeratin signal to correct for differences in load (data not shown). Graded expression along the proximaldistal axis We looked for a feature of Emx-2 expression that would be particular in pattern formation. Emx-2 mRNA was measured in different regions along the proximal-distal axis of the limb and in blastemas originating from different levels of amputation. Such blastemas regenerate different amounts of structure and show qualitative or quantitative differences, such as cellular adhesiveness, homeobox gene expression, and differential regulation of RA-activated gene expression (Crawford & Stocum, 1988; Savard et al., 1988; Brown & Brockes, 1991; Ferretti et al., 1991; Brockes, 1991; Simon & Tabin, 1993). Poly(A)‡RNA was extracted from different parts of the forelimb and hindlimb (as depicted in Figure 6(A)) and analyzed on Northern blots. Hybridization with Emx-2 revealed a 2.5 kb transcript that gradually increases in intensity in the more distal regions of both forelimb and hindlimb (Figure 6(B)). Distal blastema (midradius and ulna level) also shows a strong accumulation of Emx-2 when compared with proximal blastema (mid-humerus level). The blot was normalized to ascertain that the variations of Emx-2 expression were related to position along the limb, and not to variations in the proportion of expressing tissue in various limb parts. Emx-2 signal in distal blastemas is approximately sevenfold higher than proximal blastemas (actin as a normalizer). The same ratio is seen in proximal and distal portions of the adult limb (dlx-1 and cytokeratin 3.5 kb as normalizers). Emx-2 expression increases gradually along the proximal-distal axis of both forelimb and hindlimb as it is shown by in situ hybridization (Figure 4). The proximal-distal gradient in the intact limb is seen in Figure 4C by looking at the epidermis

505

Regulation of Emx-2 in Regeneration

Figure 4. In situ hybridization of Emx-2 in the adult newt limb and its regeneration blastema, Crosssection of a newt forelimb three weeks after amputation; the plane of amputation is labeled with a dotted line (a-a). The intact tissue if on the left side (proximal) of the line whereas the blastema tissue is on the right side (distal). (A) Bright ®eld section. (B) Autoradiograms of the same section. (C) Two times magni®cation of a selected portion of the posterior epidermis (region included between the arrowheads), to better show the proximo-distal gradient in the epidermis of the intact limb; the gradient is seen at the epidermis layer between markers m and a (left to the a-a line).

layer between markers m and a (left to the a-a line). Moreover, up-regulation of Emx-2 expression in the wound epidermis of the blastema is seen by comparing the signal left and right of the a-a line. We found similar amounts of dlx-1 messenger in the different proximal-distal regions of both forelimb and hindlimb (Figure 6(B)) as previously reported (Beauchemin & Savard, 1992). The constant level of dlx-1 expression along the limb indicates that the variations of Emx-2 expression did not result from variations in the amount of skin tissue present in each sample. The cytokeratin probe revealed three transcripts whose expression levels are differentially regulated according to position along the limb (Figure 6(B)). Although the amount of 3.5 kb transcript is constant in different regions of the limb, its seems to be upregulated in distal blastema (when compared with proximal blastema). The 3.2 kb transcript is

detectable only in the more distal regions of the limb, a pattern that is similar to the cytokeratin gene reported by Ferretti et al., 1991). Finally, the 2.8 kb transcript level decreases gradually from the proximal to the distal region of the limb, a gradient of expression that is complementary to Emx-2. The actin probe was useful to verify RNA integrity of blastemal samples, as there is no terminal differentiation in the tissue. Emx-2 and retinoic acid induced proximalization RA has a profound in¯uence on blastema development when it is injected into a newt after amputation (Niazi & Saxena, 1978; Maden, 1982). Distal blastemas (wrist level) of RA treated animals behave like proximal blastemas. Thus, RA is thought to somehow respecify the proximal-distal positional memory of the blastema. The relation-

506

Figure 5. Tissue distribution of Emx-2 in the hindlimb of adult newt. A Northern blot loaded with 5 mg of poly(A)‡RNA (absorbance at 260 nm) of various samples was serially hybridized with cytokeratin, dlx-1, Hox C6, Actin, and Emx-2. Abbreviations; Fk, ¯ank; HLt, total hindlimb; HLs, skin of the hindlimb; HLm, muscle of the hindlimb; Blt, total limb blastema, BLe, epidermis of the blastema; BLm, mesenchyme of the blastema. The numbers of right of each panel indicate the apparent length of the transcripts. The lowest band in the Emx-2 panel is speci®c to Emx-2, whereas the highest is a residual dlx-1 signal that did not wash off completely before rehybridization.

ship between proximal-distal axial position and the level of expression of Emx-2 prompted us to investigate the effect of RA on Emx-2 expression. Therefore, we used RA to proximalize the positional memory of a forelimb distal blastema and determine whether the level of Emx-2 expression is reset to a value characteristic of a more proximal blastema. Newts were amputated at the wrist level and injected either with RA or dimethylsulfoxide (DMSO) ®ve days after amputation. Blastemas were harvested seven days later, when they have reached the early bud stage. Emx-2 expression in RA treated distal blastema was compared with that of DMSO treated distal and proximal blastemas (Figure 7). Expression along the proximal-distal axis of the hindlimb was also compared. In the hindlimbs of DMSO treated newts, expression of Emx-2 is graded with higher values in more distal regions (like untreated newt Figure 6(B)). Following RA treatment, the graded expression of Emx-2 persists, but the general level of expression in all fractions is decreased by at least 50%. The same effect of RA was obtained with forelimb blastemas of different origins along the proximal-distal axis. Thus, the level of Emx-2 expression is higher in

Regulation of Emx-2 in Regeneration

Figure 6. Expression of Emx-2 along the proximaldistal axis of forelimbs and hindlimbs of adult newt. (A) Schematic representation of adult newt legs showing the code used to identify the different regions of forelimbs (F1 to F4), hindlimbs (H1 to H3) and forelimb blastema (PB, proximal blastema; DB, distal blastema). (B) Poly(A)‡RNA samples of various regions of both forelimb and hindlimb were loaded on Northern blots that were serially hybridized with Emx-2, dlx-1, cytokeratin, and actin probes. The numbers at the right of each panel indicate the apparent length of the transcripts.

distal blastema compared with proximal ones, and RA treatment on distal blastema reset the distal level of expression to a more proximal level. The normalizing probe used in comparing treated and untreated blastemas was the cytoplasmic actin, because it appeared unaffected by RA treatment, and the signal correlates well with the ethidium bromide staining of the gel.

Discussion Gene We cloned a newt homeobox-containing cDNA (Emx-2) related to the Drosophila ems gene (Dalton et al., 1989). Emx-2 is a single copy gene that encodes a 2.5 kb transcript from which we report 615 bp that include an open reading frame of 135 amino acids, starting 26 amino acids upstream of the homeodomain and ®nishing in the 30 -untranslated portion of the cDNA. The coding region is 97% identical (131 of 135) to the human EMX-2 (Simeone et al., 1992a). The degree of homology is high enough to conclude these are the newt and human version of the same gene. The high level of conservation of the protein in the phyla suggests that it plays similar functions in development and regeneration.

507

Regulation of Emx-2 in Regeneration

Figure 7. Effect of RA-treatment on Emx-2 expression in adult newt limbs. (A) Poly(A)‡RNA samples of various hindlimb regions and of proximal and distal forelimb blastemas of newts, which were treated or not with RA, were loaded on Northern blots that were serially hybridized with Emx-2, cytokeratin and actin probes. (B) Schematic representation of the ratio Emx-2/cytoplasmic actin (transcript of 1.8 kb). H1 to H3 represent different hindlimb regions (see Figure 6); BLP ‡ DMSO, proximal forelimb blastema of newts treated with DMSO; BLD ‡ RA, distal forelimb blastema of newts treated with RA; BLD ‡ DMSO, distal forelimb blastema of newts treated with DMSO. The results show that RA decreases the overall Emx-2 expression in the adult limb and its regeneration blastema.

Pattern of expression Emx-2 is mainly expressed in the epidermal components of the regeneration blastema. Its expression in the mouse limb bud is also restricted to the epidermis (Simeone et al., 1992a). Therefore, the homology between the newt and the vertebrate version of Emx-2 extends to their domain of expression and this argues in favor of the proposition that limb development and limb regeneration make use of the same genes to direct pattern formation. An Emx-2-related transcript was revealed in the muscular fraction of the limbs by reducing the stringency of hybridization (data not shown), suggesting that there is more than a single ems-like homeobox gene in the newt. This observation is in accordance with the existence of two ems-like homeobox genes (Emx-1 and Emx-2) in human and mouse (Simeone et al., 1992a,b); both genes share a

very homologous DNA-binding domain and differ in their amino and carboxy-terminal ends. Most homeobox containing genes studied to date in both insects and vertebrate are transcribed in developing tissue and show restricted expression, if any, in the adult. This is part of the general evidence that argues for their instructive role in development. The maintenance of expression of Emx-2 in adult limb and tail suggests the skin of these appendages may have some properties particular to embryonic tissue. The regeneration territories of the adult newt are believed to correspond to the embryonic ®elds that are at the origin of the formation of the body structure (Muneoka & Bryant, 1984). Therefore, sustained expression of genes of development, like Emx-2, in the adult newt may be an interesting possibility for explaining a predisposition for regeneration. For example, Emx-2 could provide the epidermis of the adult newt with developmental properties, such as the ability to form the apical ectodermal cap. Emx-2 could also provide positional information along the proximal-distal axis of the limb, to regulate the regeneration only of the missing structures. Emx-2 is expressed in the limbs and tail of the newt and not in the ¯ank; the restricted expression to regeneration territories suggests Emx-2 may be related to the regeneration potential. Analysis of its expression at the cellular level may permit delineation of the precise frontiers of the regeneration territories. It would also be interesting to investigate, in vertebrates that lose their regeneration potential during development (Dent, 1962), the time course of loss of Emx-2 expression with the loss of regenerative ability, and possibly even to maintain expression in the adult limb through appropriate manipulations (Brockes, 1992; Lourenssen & Carlone, 1993; Schilthuis et al., 1993; Burns et al., 1994; Pecorino et al., 1994). Emx-2 expression in chick limb mutants, like limbless (no AER), diplopdia-5, and talpid2 (which have a thickened AER), could also be helpful in understanding the function of the gene during limb development. Emx-2 and positional information It is interesting to relate the graded expression seen in both the forelimb and the hindlimb to the concept of positional information introduced by Lewis Wolpert (Wolpert, 1969, 1989). One particularity of this model is that positional information in two different systems could be set with the same mechanism even though the developmental outcome of each system (dependent on cell history) is different. For example, Crawford & Stocum (1988) obtained evidence that position-speci®c properties (such as cellular adhesiveness) of the blastema are homologous (or identical) in forelimb and hindlimb of the axololt. The graded expression of Emx-2 in both hindlimb and forelimb may be another example of such a mechanism. Other homeobox containing genes are expressed in the appendages of the adult newt: two distal-

508 less-like transcripts (NvHBox-4 and NvHBox-5) are found constitutively and ubiquitously in newt skin (Beauchemin & Savard, 1992); FH-1 is also found in the skin, but only in the forelimbs and not the hindlimb (Tabin, 1989); Hox C6, Hox A11, Hox A2, Hox A3, Hox-C10 and Msx-1 are present in the mesenchymal and skin fractions of limb, tail, and their corresponding blastemas (Savard et al., 1988; Beauchemin & Savard, 1993; Simon & Tabin, 1993; Beauchemin et al., 1994; Crews et al., 1995). Two other homeobox genes (Hox D10 and Hox D11) are expressed only in the mesenchymal fraction of the regeneration blastema (Brown & Brockes, 1991; Simon & Tabin, 1993). It is interesting that Hox A11, Hox C6, Hox D10, and Hox D11 show a higher level of expression in more proximal regions of the limb, whereas Emx-2 shows an opposite gradient, meaning that the epidermal and mesenchymal compartments of the limb blastema are expressing different combinations of homeobox containing genes, depending on their location in the limb. This, we suspect, could be related to the identi®cation of cell position in the appendage. Homeobox containing genes are suspected to play important functions in the patterning of vertebrate limbs, and craniofacial structures by providing an ordered molecular system of positional values (IzpisuÂaBelmonte et al., 1991; Kessel & Gruss, 1991; Nohno et al., 1991; Kessel, 1992; Marshall et al., 1993). Homeobox containing gene expression in the adult newt limb, and its regeneration blastema, suggests that the combination of genes used during limb development may still be active or reactivated during regeneration in the adult newt. Because the covering epidermis appears to provide only a permissive in¯uence on blastema development (Stocum & Dearlove, 1972; Tassava & Mescher, 1975), it is unlikely that a homeobox combination found in the epidermis would be instructive to the regeneration blastema. In turn, the identity of the underlying mesenchyme might induce the overlying epidermis to acquire expression of speci®c homeobox genes just as it can induce speci®c ectodermal derived structures (Hamburgh, 1970; Saunders, 1980). Therefore, signaling from mesenchyme to epidermis may induce ectodermal phenotypes that are important in positional speci®cation. The hypothesis that Emx-2 level of expression is related to cell position in the limb, and thus participates in blastema determination, can be certi®ed by changing the blastema determination and analyzing whether Emx-2 expression is respeci®ed to a value characteristic of the new blastemal fate. It has been possible to proximalize both the blastemal fate and the level of expression of Emx-2 using RA treatment. A similar type of proximal regulation in RA treated blastema was observed with Hox D10 and Hox A13 (Simon & Tabin, 1993; Gardiner et al., 1995). Knowing the important role of homeobox-containing genes in the determination of positional information during development, it is conceivable that homeobox-containing

Regulation of Emx-2 in Regeneration

genes would act by specifying position along the limb to ensure that only portions distal to the amputation plan are regenerated. Indeed, it was proposed that the wound epidermis which migrates over the amputation surface serves as the distal boundary during regeneration (Maden, 1977; Repesh & Oberbriller, 1978; Stocum, 1984). RA receptors (RARs) are strong candidates to be the initial targets in altering positional memory of the blastema. The results of Pecorino et al. (1996a) support this hypothesis since the activation of the RAR d2 isoform speci®cally mediates cell proximalization. On the other hand, the four isoforms of RAR analyzed to date do not shown marked differences in expression either along the proximaldistal axis of intact limbs or between proximal and distal blastemas (GigueÁre et al., 1989; Ragsdale et al., 1989, 1992a,b; Ragsdale & Brockes, 1991; Hill et al., 1993). Therefore, we do not exclude the participation of secondary factors that would interact differently with the RAR pathways. The expression of transcription factors, like homeobox-containing genes, could be candidate for such a mechanism.

Materials and Methods Animals and treatment Adult N. viridescens were purchased from C. D. Sullivan Co. Inc. (Nashville, Tennessee). Anesthesia and surgical procedures were as reported (Savard et al., 1988). Blastemas were harvested at the mid-bud stage. When needed, blastema epidermis was mechanically dissected of the mesenchyme in a calcium and magnesium free PBS buffer (20 minutes of incubation). Proximalization of distal blastemas was done ®ve days postamputation with a single intraperitoneal injection of 20 ml of a solution of DMSO-RA (15 mg/ml). Control animals were injected with DMSO. The blastemas were collected seven days after treatment. Cloning, PCR-RACE, and sequencing The p027 clone was isolated by screening a newt cDNA library with a redundant oligonucleotide recognizing a highly conserved region of the homeobox sequence. Screening was as described (Beauchemin & Savard, 1992). cDNA extensions in the 50 region of the gene were cloned by PCR-RACE (Frohman et al., 1988; Frohman, 1990; Jain et al., 1992). Brie¯y, 1 mg of forelimb poly(A)‡RNA was oligo(dT) primed to synthesize cDNA; the mix was subsequently ®ltrated on a Centricon 100 spun-column, and poly(A)-tailed with 17 units of T4 deoxynucleotidyl transferase (Pharmacia) for 15 minutes at 37 C, in 200 ml of DNA-tailing buffer (0.1 M potassium-cacodylate (pH 7.2), 2 mM cobalt chloride, 0.2 mM dithiothreitol). A ®rst series (30 rounds) of PCR ampli®cation was done with 10 ml of the previous reaction and 150 ng of both oligonucleotides poly(T) (50 -AAGGATCCGTCGACAT-CGATAATACGAC(T)17-30 ), and PUY-40 (50 -CTATTCGCCACCTGTTGA-30 ; position 341 to 358 of the reported sequence in Figure 2) in 50 ml; of PCR reaction buffer (67 mM Tris (pH 8.8), 6.7 mM magnesium chloride, 170 mg/ml bovine serum albumin, 17 mM ammonium sulfate, 1.5 mM deoxynucleotide

509

Regulation of Emx-2 in Regeneration triphosphate, and 10% DMSO) containing 2.5 units Taq DNA polymerase (Pharmacia). The PCR program followed these conditions: a hot start (95 C for seven minutes, Taq addition, 75 C for two minutes, 45 C for two minutes, and 72 C for 40 minutes), and 30 regular cycles (94 C for 45 seconds, 50 C for 25 seconds, 72 C for three minutes with an extension cycle of ®ve seconds). A second series (30 rounds) of PCR ampli®cation was done with 1 ml of the ®rst ampli®cation and 150 ng of both oligonucleotides Ri (50 -GACATCGATAATACGAC-30 ; poly(T)-complementary sequence) and PUY-41 (50 -GTGCCCTTCTTCTTCTGC-30 ; position 315 to 336 of the reported sequence in Figure 2). The ®nal products of ampli®cation were cloned in the PCR-II vector (TA-cloning kit, In Vitrogen Corporation). Positive clones were detected by colony hybridization using an Emx-2 speci®c oligonucleotide located in the homeobox. Nucleotide sequence determination was performed by the method of Sanger, using the kits and methods provided by United States Biochemical Corporation and Pharmacia. RNA isolation, Northern blots and in situ hybridization RNA was extracted as described by Chomczynski & Sacchi (1987). Poly(A)‡RNA was puri®ed on an oligo(dT)-cellulose column following the procedure provided by the supplier (Boehringer Mannheim). About 5 mg of poly(A)‡RNA (evaluated with the optical density at 260 nm) was run onto an agarose-formaldehyde gel (Maniatis et al., 1982) and transferred onto Genescreen membranes according to the supplier (New England Nuclear). The hybridization procedure and normalizing probes were as previously described (Beauchemin & Savard, 1992). Autoradiograms were scanned with an image analyzer RAS-1000 (Amersham) to evaluate band intensity. For in situ hybridization, regenerating limbs were frozen in Optimal Cutting Temperature compound (OCT), sectioned and processed as described (Del Rio-Tsonis et al., 1995). Brie¯y, sections were ®xed in 4% paraformaldehyde, rinsed in PBS, treated with triethanolamine, washed with 2 sodium citrate buffer (SSC), and dehydrated by ethanol series (50% to 100%). Slides were allowed to air dry and were then incubated overnight at 55 C with either the sense or antisense 35S-labelled Emx-2 probes. The next day, the slides were rinsed in 2 SSC, then washed in a solution containing 50% formamide, 1 SSC, 0.1% b-mercaptoethanol at 52 C followed by RNase treatment, and ®nally washed with 0.1 SSC and 0.1% b-mercaptoethanol. Sections were dehydrated and air dried so that they could be dipped in photographic emulsion and exposed for two weeks. Slides were counterstained with either hematoxylin-eosin or van Gieson's stain.

Acknowledgments We thank Ms Jean Daly for helping in the preparation of the manuscript. M.B. had a studentship from the Georges PheÂnix Foundation and P.S. had a scholarship from the Medical Research Council of Canada (MRC) and the ``Fonds de la Recherche en Sante du QueÂbec. This work was funded by the MRC (MT-10664).

References Beauchemin, M. & Savard, P. (1992). Two Distal-less related homeobox-containing genes expressed in regeneration blastemas of the newt. Dev. Biol. 154, 55 ±65. Beauchemin, M. & Savard, P. (1993). Expression of ®ve homeobox genes in the adult newt appendages and regeneration blastemas. In Limb Development and Regeneration (Fallon, J. F., Goetinck, P. F., Kelly, R. O. & Stocum, D. L., eds), pp. 41 ± 50, Wiley-Liss, Inc., USA. Beauchemin, M., Noiseux, N., Tremblay, M. & Savard, P. (1994). Expression of Hox A11 in the limb and the regeneration blastema of adult newt. Int. J. Dev. Biol. 38, 641± 649. Brockes, J. P. (1992). Introduction of a retinoid reporter gene in urodele limb blastema. Proc. Natl Acad. Sci. USA, 89, 11386± 11390. Brockes, J. P. (1996). Retinoid signalling and retinoid receptors in amphibian limb regeneration. Biochem. Soc. Symp. 62, 137±142. Brockes, J. P. (1997). Amphibian limb regeneration: rebuilding a complex structure. Science, 276, 81 ± 87. Brown, R. & Brockes, J. P. (1991). Identi®cation and expression of a regeneration-speci®c homeobox gene in the newt limb blastema. Development, 111, 489± 496. Burns, J. C., Matsubara, T., Lozinski, G., Yee, J.-K., Friedman, T., Washabaugh, C. H. & Tsonis, P. A. (1994). Pantropic retroviral vector-mediated gene transfer, integration, and expression in cultured newt limb cells. Dev. Biol. 165, 285± 289. Chomczynski, P. & Sacchi, N. (1987). Single-step method of RNA isolation by acid guanidium thiocyanatephenol-chloroform extraction. Anal. Biochem. 162, 156± 159. Crawford, K. & Stocum, D. L. (1988). Retinoic acid coordinately proximalizes regenerate pattern and blastema differential af®nity in axolotl limbs. Development, 102, 687± 698. Crews, L., Gates, P. B., Brown, R., Joliot, A., Foley, C., Brockes, J. P. & Gann, A. A. (1995). Expression and activity of the newt Msx-1 gene in relation to limb regeneration. Proc. Roy. Soc. Lond. 259, 161± 171. Dalton, D., Chadwick, R. & McGinnis, W. (1989). Expression and embryonic function of empty spaces: a Drosophila homeobox gene with two patterning functions on the anterior-posterior axis of the embryo. Genes Dev. 3, 1940± 1956. Davies, C. A., Holmyard, D. P., Millen, K. J. & Joyner, A. L. (1991). Examining pattern formation in mouse, chicken and frog embryos with an Enspeci®c antiserum. Development, 111, 287± 298. Del, Rio-Tsonis K., Washabaugh, C. H. & Tsonis, P. A. (1995). Expression of pax-6 during urodele eye development and lens regeneration. Proc. Natl Acad. Sci. USA, 92, 5092± 5096. Dent, J. N. (1962). Limb regeneration in larvae and metamorphosing individuals of the South African clawed toad. J. Morphol. 110, 61± 78. DolleÂ, P., IzpisuÂa-Belmonte, J.-C., Falkenstein, H., Renucci, A. & Duboule, D. (1989). Coordinate expression of the murine Hox-5 complex homeobox-containing genes during limb pattern formation. Nature, 342, 767± 772. Ferretti, P. & Brockes, J. P. (1991). Cell origin and identity in limb regeneration and development. GLIA, 4, 214± 224.

510 Ferretti, P., Brockes, J. P. & Brown, R. (1991). A newt type II keratin restricted to normal and regenerating limbs and tails is responsive to retinoic acid. Development, 111, 497± 507. Frohman, M. A. (1990). RACE: Rapid ampli®cation of cDNA ends. In PCR Protocols (Innis, M. A., Gelfand, D. H., Snincky, J. J. & White, T. J., eds), pp. 28 ± 38, Academic Press, San Diego, CA. Frohman, M. A., Dush, M. K. & Martin, G. R. (1988). Rapid production of full-length cDNAs from rare transcripts: ampli®cation using a single genespeci®c oligonucleotide primer. Proc. Natl Acad. Sci. USA, 85, 8998± 9002. Gann, A. A. F., Gates, P. B., Stark, D. & Brockes, J. P. (1996). Receptor isoform speci®city in a cellular response to retinoic acid. Proc. Roy. Soc. London Ser. B, 263, 729± 734. Gardiner, D. M., Blumberg, B., Komine, Y. & Bryant, S. V. (1995). Regulation of HoxA expression in developing and regenerating and axolotl limbs. Development, 121, 1731 ±1741. GigueÁre, V., Ong, E. S., Evans, R. M. & Tabin, C. J. (1989). Spatial and temporal expression of the retinoic acid receptor in the regenerating amphibian limb. Nature, 337, 566±569. GuyeÂnot, E., Dinichert-Favarger, J. & Galland, M. (1948). L'exploration du territoire de la patte anteÂrieure du Triton. Rev. Suisse Zool. 55, 1 ± 120. Hamburgh, M. (1970). Theories of Differentiation, Elsevier, New York. Hill, D. S., Ragsdale, C. W., Jr & Brockes, J. P. (1993). Isoform-speci®c immunological detection of newt retinoic acid receptor d1 in normal and regenerating limbs. Development, 117, 937± 945. IzpisuÂa-Belmonte, J.-C., Tickle, C., DolleÂ, P., Wolpert, L. & Duboule, D. (1991). Expression of the homeobox Hox-4 genes and the speci®cation of position in chick wing development. Nature, 350, 585± 589. IzpisuÂa-Belmonte, J.-C., Ede, D. A., Tickle, C. & Duboule, D. (1992). The mis-expression of posterior Hox-4 genes in talpid (ta3) mutant wings correlates with the absence of anteroposterior polarity. Development, 114, 959± 963. Jain, R., Gomer, R. H. & Murtagh, J. J. (1992). Increasing speci®city from the PCR-RACE technique. Biotechniques, 12, 58 ± 59. Kessel, M. (1992). Respeci®cation of vertebral identities by retinoic acid. Development, 115, 487± 501. Kessel, M. & Gruss, P. (1991). Homeotic transformations of murine vertebrae and concomitant alteration of Hox codes induced by retinoic acid. Cell, 67, 89 ± 104. Kiortsis, V. (1953). PotentialiteÂs du territoire patte chez le Triton. Rev. Suisse Zool. 60, 301± 410. Lo, D. C., Allen, F. & Brockes, J. P. (1993). Reversal of muscle differentiation during urodele limb regeneration. Proc. Natl Acad. Sci. USA, 90, 7230± 7234. Lourenssen, S. & Carlone, R. (1993). Transfection of adult newt limb blastema cells in vivo: increased ef®ciency with direct injection of plasmid DNA. In Limb Development and Regeneration (Fallon, J. F., Goetinck, P. F., Kelly, R. O. & Stocum, D. L., eds), pp. 193± 201, Wiley-Liss, Inc., USA. Maden, M. (1977). The regeneration of positional information in the amphibian limb. J. Theoret. Biol. 69, 735± 753. Maden, M. (1982). Vitamin A and pattern formation in the regenerating limb. Nature, 295, 672± 675.

Regulation of Emx-2 in Regeneration Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982). In Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Marshall, H., Nonchev, S., Sham, M. H., Muchamore, I., Lumsden, A. & Krumlauf, R. (1993). Retinoic acid alters hindbrain Hox code and induces transformation of rhombomeres 2/3 into a 4/5 identity. Nature, 360, 737± 741. Mathisen, P. M. & Miller, L. (1987). Thyroid hormone induction of keratin genes: a two-step activation of gene expression during development. Genes Dev. 1, 1107± 1117. McGinnis, W. & Krumlauf, R. (1992). Homeobox genes and axial patterning. Cell 68, 283± 302. Miyamoto, N., Yoshida, M., Kuratani, S., Matsuo, I. & Aizawa, S. (1997). Defects of urogenital development in mice lacking Emx2. Development, 124, 1653± 1664. Mohun, T. J., Brennan, S., Dathan, N., Fairman, S. & Gurdon, J. B. (1984). Cell-type-speci®c activation of actin genes in the early amphibian embryo. Nature, 311, 716± 721. Muneoka, K. & Bryant, S. V. (1984). Cellular contribution to supernumerary limbs resulting from the interaction between developing and regenerating tissues in the axololt. Dev. Biol. 105, 179± 187. Muneoka, K. & Sassoon, D. (1992). Molecular aspects of regeneration in developing vertebrate limbs. Dev. Biol. 152, 37 ± 49. Niazi, A. A. & Saxena, S. (1978). Abnormal hindlimb regeneration in tadpoles of the toad, Bufo andersoni, exposed to excess vitamin A. Folia Biol. (Krakow), 26, 3 ± 11. Nohno, T., Noji, S., Koyama, E., Ohyama, K., Myokai, F., Kuroiwa, A., Saito, T. & Taniguchi, S. (1991). Involvement of the Chox-4 chicken homeobox genes in determination of anteroposterior axial polarity during limb development. Cell, 64, 1197± 1205. Oliver, G., Sidell, N., Fiske, W., Heinzmann, C., Mohandas, T., Sparkes, R. S. & De Robertis, E. M. (1989). Complementary homeo protein gradients in developing limb buds. Genes Dev. 3, 641±650. Pecorino, L. T., Lo, D. C. & Brockes, J. P. (1994). Isoform-speci®c induction of a retinoid-responsive antigen after biolistic transfection of chimaeric retinoic acid/thyroid hormone receptors into a regenerating limb. Development, 120, 325± 333. Pecorino, L. T., Entwistle, A. & Brockes, J. P. (1996a). Activation of a single retinoic acid receptor isoform mediates proximodistal respeci®cation. Curr. Biol. 6, 563± 569. Pecorino, L. T., Brockes, J. P. & Entwistle, A. (1996b). Semi-automated positional analysis using laser scanning microscopy of cells transfected in a regenerating newt limb. J. Histochem. Cytochem. 44, 559± 569. Peterson, R. L., Papenbrock, T., Davda, M. M. & Awgulewitsch, A. (1994). The murine HoxC cluster contains ®ve neighboring AbdB-related Hox genes that show unique spatially coordinated expression in posterior embryonic subregions. Mech. Dev. 47, 253± 260. Ragsdale, C. W. & Brockes, J. P. (1991). Retinoic receptors and vertebrate limb morphogenesis. In Nuclear Hormone Receptors (Parker, M. G., ed.), pp. 269± 295, Academic Press, London. Ragsdale, C. W., Petkovich, M., Gates, P. B., Chambon, P. & Brockes, J. P. (1989). Identi®cation of a novel

Regulation of Emx-2 in Regeneration retinoic acid receptor in regenerative tissues of the newt. Nature, 341, 654± 657. Ragsdale, C. W., Gates, P. B. & Brockes, J. P. (1992a). Identi®cation and expression pattern of a second isoform of the newt alpha retinoic acid receptor. Nucl. Acid Res. 20, 5851. Ragsdale, C. W., Gates, P. B., Hill, D. S. & Brockes, J. P. (1992b). Delta retinoic acid receptor isoform d1 is distinguished by its exceptional N-terminal sequence and abundance in the limb regeneration blastema. Mech. Dev. 40, 99± 112. Repesh, L. A. & Oberbriller, J. L. (1978). Scanning electron microscopy of epidermal cell migration in wound healing during limb regeneration in the adult newt, Notophthalmus viridescens. Am. J. Anat. 151, 539± 556. Ros, M. A., Lyons, G., Kosher, R. A., Upholt, W. B., Coelho, C. N. D. & Fallon, J. F. (1992). Apical ridge dependent and independent mesodermal domains of GHox-7 and GHox-8 expression in chick limb buds. Development, 116, 811± 818. Saunders, J. W., Jr (1980). Developmental Biology, Macmillan, New York. Savard, P. & Tremblay, M. (1995). Differential regulation of Hox C6 in the appendages of adult urodeles and anurans. J. Mol. Biol. 249, 879± 889. Savard, P., Gates, P. B. & Brockes, J. P. (1988). Position dependent expression of a homeobox gene transcript in relation to amphibian limb regeneration. EMBO J. 7, 4275± 4282. Schilthuis, J. G., Gann, A. A. F. & Brockes, J. P. (1993). Chimeric retinoic acid-thyroid hormone receptors implicate RAR-a1 as mediating growth inhibition by retinoic acid. EMBO J. 12, 3459± 3466. Scott, M. P., Tamkun, J. W. & Hartzell, G. W., III (1989). The structure and function of the homeodomain. Biochem. Biophys. Acta, 989, 25 ± 48. Simeone, A., Gulisano, M., Acampora, D., Stornaiuolo, A., Rambaldi, M. & Boncinelli, E. (1992a). Two vertebrate homeobox genes related to the Drosophila empty spiracles gene are expressed in the embryonic cerebral cortex. EMBO J. 11, 2541± 2550.

511 Simeone, A., Acampora, D., Gulisano, M., Stornaiuolo, A. & Boncinelli, E. (1992b). Nested expression domains of four homeobox genes in developing rostral brain. Nature, 358, 687±690. Simon, H.-G. & Tabin, C. J. (1993). Analysis of Hox-4.5 and Hox-3.6 expression during newt limb regeneration: differential regulation of paralogous Hox genes suggest different roles for members of different Hox clusters. Development, 117, 1397± 1407. Stocum, D. L. (1984). The urodele limb regeneration blastema: determination and organization of the morphogenetic ®eld. Differentiation, 27, 13 ± 28. Stocum, D. L. & Dearlove, G. E. (1972). Epidermal-mesodermal interaction during morphogenesis of the limb regeneration blastemas in larval salamanders. J. Exp. Zool. 181, 49 ± 61. Tabin, C. J. (1989). Isolation of potential vertebrate limbidentity genes. Development, 105, 813± 820. Tanaka, E. M., Gann, A. A., Gates, P. B. & Brockes, J. P. (1997). Newt myotubes reenter the cell cycle by phosphorylation of the retinoblastoma protein. J. Cell. Biol. 136, 155± 165. Tassava, R. A. & Mescher, A. L. (1975). The roles of injury, nerves and the wound epidermis during the initiation of amphibian limb regeneration. Differentiation, 4, 23 ±24. Tsonis, P. A. (1996). Limb Regeneration, Cambridge University Press, Cambridge, UK. Wallace, H. (1981). Vertebrate Limb Regeneration, John Wiley & Sons, Chichester. Wolpert, L. (1969). Positional information and the spatial pattern of cellular differentiation. J. Theoret. Biol. 25, 1 ±47. Wolpert, L. (1989). Positional information revisited. Development, 109(suppl.), 3 ± 12. Yokouchi, Y., Sasaki, H. & Kuroiwa, A. (1991). Homeobox gene expression correlated with the bifurcation of limb cartilage development. Nature, 353, 443± 445. Yoshida, M., Suda, Y., Matsuo, I., Miyamoto, N., Takeda, N., Kuratani, S. & Aizawa, S. (1997). Emx1 and Emx2 functions in development of dorsal telencephalon. Development, 124, 101±111.

Edited by J. Karn (Received 24 November 1997; received in revised form 9 March 1998; accepted 16 March 1998)