Expression patterns of bone morphogenetic proteins (Bmps) in the developing mouse tooth suggest roles in morphogenesis and cell differentiationкод для вставкиСкачать
DEVELOPMENTAL DYNAMICS 210:355–370 (1997) Keratin 8 and 18 Expression in Mesenchymal Progenitor Cells of Regenerating Limbs Is Associated With Cell Proliferation and Differentiation JONATHAN P. CORCORAN AND PATRIZIA FERRETTI* Developmental Biology Unit, Institute of Child Health, University College London, London, United Kingdom ABSTRACT Keratins are considered markers of epithelial differentiation. In lower vertebrates, however, immunoreactivity for keratin 8 and 18 has been reported in nonepithelial cells, particularly in mesenchymal progenitor cells of regenerating complex body structures. To confirm that such reactivity does indeed reflect keratin expression and to investigate their possible role in regeneration, we have isolated clones coding for the newt homologues of keratin 8 and 18 (NvK8 and NvK18, respectively) and studied their distribution and changes in their expression following experimental manipulations. Analysis of NvK8 and NvK18 transcripts confirms that K8 and K18 are expressed in the blastemal cells of regenerating newt limbs and that their expression is first observed 3-5 days after amputation, when the blastemal cells start to proliferate under the influence of the nerve, whose presence is essential for regeneration to proceed. In contrast, no induction of these keratins is observed following amputation of a larval limb at a stage when organogenesis is proceeding in a nerve-independent manner. To establish whether there is a causal relationship between keratin expression and cell proliferation in the adult limb blastema, we have investigated whether their expression is nerve-dependent and whether suppression of their expression in cultured blastemal cells affects cell division and differentiation. Analysis of keratins in denervated limbs demonstrates that the nerve is not necessary to induce their expression. However, treatment of cultured blastemal cells with K8 and K18 anti-sense oligonucleotides significantly decreases DNA synthesis and induces changes in cell morphology, suggesting that expression of these keratins during regeneration may be necessary for the maintenance of the undifferentiated and proliferative state of blastemal cells. Dev. Dyn. 1997;210:355–370. r 1997 Wiley-Liss, Inc. Key words: differentiation; limb; keratin; regeneration; proliferation; urodele amphibians INTRODUCTION Urodele amphibians, such as newts and axolotls, can regenerate their limbs as adults through a process r 1997 WILEY-LISS, INC. called epimorphic regeneration. In the course of this regenerative process, a mound of undifferentiated mesenchymal progenitors cells, the blastemal cells, accumulate under a specialized wound epidermis, which rapidly covers the stump surface following amputation. After this accumulation phase, the blastemal cells start to proliferate. Up to 2–3 weeks (Thornton, 1968) after amputation, before the regeneration blastema has become palette-shaped and differentiation and morphogenesis begin, limb denervation results in the arrest of DNA synthesis and consequent inhibition of regeneration (Singer, 1952; Thornton, 1968). However, when the regenerating limb is denervated after reaching a critical size (from the late bud stage onward), a complete limb, although smaller than the amputated one, will form (Schotté and Butler, 1944; Singer, 1948). Nervedependent growth control of regeneration is established as a consequence of innervation. In fact, whereas the presence of the nerve is required during regeneration of limbs that have been innervated, regeneration of a developing limb bud can occur before it becomes innervated. A distinction between the progenitor cells of developing and regenerating limb buds is highlighted by their different content of intermediate filaments. Although both populations express the mesenchymal intermediate filament vimentin, reactivity with the monoclonal antibody (mAb) 22/18 and mAbs against human simple epithelial keratins 8 and 18 (K8 and K18) is observed only in regenerating limbs (Fekete and Brockes, 1987; Ferretti et al., 1989). This apparent expression of K8 and K18 in mesenchymal cells is puzzling because cytokeratins, obligatory heteropolymers encoded by two large multigene families, are usually considered markers of epithelial differentiation and are normally expressed in glandular epithelia. In lower vertebrates, keratin immunoreactivity has also been reported in some nonepithelial tissues that grow throughout life and have significant regenerative ability, such as optic nerve and fins in fish (Giordano et al., 1989; Ferretti et al., in preparation), and in regeneration blastemas of Grant sponsor: Wellcome Trust. Dr. Corcoran’s current address is Developmental Biology Research Center, The Randall Institute, King’s College London, 26-29 Drury Lane, London WC2B 5RL, United Kingdom. *Correspondence to: Dr. Patrizia Ferretti, Developmental Biology Unit, Institute of Child Health, UCL, 30 Guilford Street, London WC1N 1EH, United Kingdom. E-mail: firstname.lastname@example.org Received 10 July 1997; Accepted 19 August 1997 356 CORCORAN AND FERRETTI different organs in amphibians (Ferretti et al., 1989; Ferretti and Ghosh, 1997). Furthermore, K8 and K18 are present in the oocyte and are among the first genes to be activated during embryogenesis when the cells are still pluripotent and no differentiation is apparent. The role of this keratin pair in cell function is still elusive, but the view that these and other intermediate filaments are only needed to stabilize the cell structure has been challenged by the fact that their assembly and disassembly has turned out to be much more dynamic than initially believed and that changes in their expression can be associated with changes in cell physiology. For example, depletion of these keratins in early Xenopus embryos results in gastrulation abnormalities (Heasman et al., 1992), activation of K8 and 18 in certain tumors may have a role in their increased migratory and invasive ability (Chu et al., 1993; Oshima et al., 1996), lack of K8 in transgenic mice is lethal (Baribault et al., 1993), and mutations in human K18 appear to be associated with cryptogenic cirrhosis (Ku et al., 1995, 1997). Therefore, the fact that simple epithelial keratins are expressed in many instances in nonepithelial tissues following amputation may indeed reflect a significant role of these molecules in the regenerative process. This possibility has yet to be thoroughly investigated, and such studies might not only help to elucidate the mechanisms underlying the regenerative process, but also to reveal yet unidentified roles of simple epithelial keratins in cell function. The regenerating limb offers an important model for such a study, because it is so far the best characterized of the regenerating systems in lower vertebrates and in culture (Géraudie and Ferretti, 1997). As mentioned above, in the newt, Notophthalmus viridescens, reactivity with antibodies to human K8 and K18 has been observed in the blastema of the regenerating adult limb but not in progenitor cells of normally developing limb buds in a related species, Pleurodeles waltl, although various embryonic tissues were strongly stained (Ferretti et al., 1989). Therefore, it was important to demonstrate conclusively that K8 and K18 are the proteins upregulated following amputation and that they are not expressed in the developing limb buds in the same species. Furthermore, we wished to establish whether expression of newt K8 and K18 could be induced following amputation of limb buds in N. viridescens embryos or whether, like 22/18, their expression was induced only after the transition from the nerve-dependent to the nerve-independent mode of regeneration. Finally, we wished to investigate whether there is a causal relationship between expression of these keratins in blastemal cells and the regenerative capability of the urodele limb, particularly in relation to nerve dependency, because of the apparent difference in keratin expression in embryonic limb buds and regeneration blastemas. Therefore, we have isolated cDNA clones coding for the newt keratin homologues of mammalian K8 and K18 and studied the distribution of these transcripts both in developing and regenerating limbs under different experimental conditions for two purposes. First, to confirm that the immunoreactivity detected in blastemal cells with anti-keratin mAbs was indeed due to the expression of simple epithelium keratins. Second, to assess a possible role for these proteins in the regenerative process. We show here that these keratins are expressed in the blastema mesenchyme and in cultured limb cells, but not in regenerating limb buds before they become innervated, and that, although they are first expressed in the regeneration blastema at the onset of cell proliferation, their induction is not dependent on the presence of innervation. We have also explored the possibility that newt K8 and K18 might be permissive for cell proliferation in the blastema by downregulating their expression in culture using anti-sense oligonucleotides. We show that this results in inhibition of proliferation and changes in cell morphology that are consistent with a role of simple epithelial keratins in maintaining the blastemal phenotype. RESULTS Isolation of NvK8 and NvK18 cDNA Clones Eight of the plaques identified by screening an expression cDNA library from cultured blastemal cell with the mAb LP1K, which recognizes a single protein of apparent molecular weight 52 kD in these cultures (Ferretti et al., 1989), were purified. Seven of the eight clones isolated shared the same PstI and SacI restriction sites. The three longest overlapping clones (1700, 1440, and 1250 bp) were selected for sequence analysis (Fig. 1A). All these clones extended into the 38 end untranslated region, and one of them extended toward the 58 end region 225 bp further than the others but did not contain a methionine initiation codon. Of the plaques identified by hybridization with the human K18 probe, five were isolated and they all shared the same BglII and PstI restriction sites. Three of these clones were used for further analysis, and the longest clone (which contains the two shorter ones) was 2400 bp (Fig. 1B). DNA Sequence and Deduced Amino Acid Sequence of cDNA Clones The basic structure of an intermediate filament protein that comprises a linear N-terminus (head), an a-helical central portion (rod), and a linear C-terminus (tail) is shown in Figure 1C. The nucleotide and deduced amino acid sequence of the overlapping clones isolated by LP1K screening is shown in Figure 2A. These clones code for a type II cytokeratin, which was named N. viridescens keratin 8 (NvK8), since sequence comparison with other proteins shows that the highest percentage of NvK8 homology is with proteins of the type II keratin gene family, particularly with the simple epithelia keratin 8 (Fig. 2B). The percentage of amino acid identity in the entire rod domain of NvK8 is much higher with goldfish, Xenopus, and mouse K8 (83%, 87%, and 81%, respectively) than with the previously cloned type II newt keratin, NvKII (69%) (Ferretti et MESENCHYMAL PROGENITOR PROLIFERATION 357 initiation codon, suggests that NvK8 sequence is not complete and is probably missing 10 amino acids. Partial sequencing of the clones isolated using the human K18 probe shows that they encode the newt homologue, N. viridescens keratin 18 (NvK18), of K18 (Fig. 3). The sequenced portion of the NvK18 rod domain, which includes most helix 1B (82 of 101 amino acids) and 11 amino acids of the linker joining helix 1B and helix 2, contains the motif DNARPAADDFR, which is highly conserved in type I keratins. Comparison analysis shows that this region of NvK18 shares 70% and 69% amino acid identity with Xenopus and mouse K18, respectively (Singer et al., 1986; LaFlamme et al., 1988), 47% identity with human keratin 16 (HK16; Rosenberg et al., 1988), and 57% identity with human keratin 17 (HK17; RayChaudhury et al., 1986). The carboxyl tail of NvK18 has high amino acid identity with Xenopus (72%). However, in contrast to the high degree of conservation with mammalian type I keratins in the rod domain, the amino acid identity of NvK18 tail with mouse K18 is 55% and that with HK16 and HK17 only 10% and 28%, respectively. However, there is a conserved motif at the end of the C-terminus that is also found in the simple epithelia type I keratin HK17 and in K8s of amphibians, fish, and mammals (Figs. 2B, 3). Tissue Distribution of the NvK8 and NvK18 Transcripts Fig. 1. Restriction maps of newt clones isolated using the mAb LP1K (A) and a human cDNA K18 probe (B) as well as a schematic representation (C) of the structure of keratin proteins showing the target regions of the anti-sense oligomers used in this study. (A) Three overlapping clones coding for the newt keratin 8 (NvK8). The 300 nucleotide NvK8 PstI-PstI subclone used to make riboprobes is shown. (B) Three overlapping clones coding for the newt keratin 18 (NvK18). The 290 nucleotide NvK18 BglII-EcoRI subclone used to make riboprobes is shown. The dotted lines in A and B indicate 38 untranslated sequences. The double bar in B indicates the beginning of the coding region. (C) Highly schematic representation of a keratin. The hatched boxes indicate the a-helical portions of the central rod domain (helix 1A, helix 1B, and helix 2), which are joined by linear linkers (thin lines), L1, and L12. The straight lines flanking the rod represent the linear N-terminus region (head) and C-terminus region (tail). The regions against which the anti-sense oligomers to NvK8 and NvK18 are targeted and their sequences are shown. al., 1991). In the head and tail, a high percentage of identity with other K8s is observed in the regions flanking the rod domain and in motifs toward the end of both termini. The motif at the end of the C-terminus (Fig. 2A) is found in a similar position in all K8 proteins, in several type I keratins, and in other nonepithelial intermediate filaments such as vimentin (Quax et al., 1983). The tail of NvK8, as that of simple epithelia keratins from other lower vertebrates (Franz and Franke, 1986; Giordano et al., 1989; Charlebois et al., 1990), is longer than its mammalian counterpart and is rich in glycine. Five repeats of the type GGS/Y are present. The sequence of NvK8 at the N-terminus is 10 amino acids shorter than that of other K8s; this, together with the fact that we did not find a methionine To obtain an initial indication of the transcriptional regulation of NvK8 and NvK18, Northern analysis was performed using RNA from unamputated forelimbs, forelimb blastemas, liver, intestine, and esophagus. Both NvK8 and NvK18 probes detected single transcripts of approximately 2.1 and 1.4 Kb, respectively (Fig. 4A,B), in liver, proximal forelimb blastemas, and intestine. No signal was detected with either probe in normal limbs and esophagus. Thus, it appears that there is a single functional copy of both NvK8 and NvK18 and that these transcripts are upregulated in the regenerate. Further analysis of the expression of NvK8 at different axial levels in normal and regenerating limbs and tails was carried out by RNAase protection because of the higher sensitivity of the method (Fig. 4C). This analysis shows that in the unamputated limb, the NvK8 transcript is hardly detectable in the arm but clearly visible in the hand and fingertips. The levels of the NvK8 transcript, however, are significantly higher in the regenerate, but no difference between proximal and distal blastemas is observed. A significant increase in NvK8 mRNA is also observed in newt tail blastemas compared with normal tail. These results further indicate that upregulation of NvK8 is associated with the regenerative process and, as previously suggested by immunocytochemistry, that the fingertips are blastemalike. To establish whether NvK8 and NvK18 transcripts are restricted to the mesenchyme and not expressed Fig. 2. Nucleotide and amino acid sequence of NvK8. GenBank Accession Number 136454. (A) The nucleotide and predicted amino acid sequence of the composite NvK8 cDNAs is shown. There is no initiation methionine, suggesting that the available cDNA is not full length; the comparison analysis in B indicates the lack of 10 amino acids. Two amino acid motifs, TYRKLLEGE, which marks the end of the helical central rod domain, and DGKVTSES, which is similar to a motif found in the C-terminus of K8 in other species, are underlined. (B) Comparison of the amino acid sequence of NvK8 with that of Xenopus (XK8), mouse (MK8), bovine (BK8), and goldfish (gfK8) keratin 8 and with the newt type II keratin, NvKII. Identical amino acids are in bold. The rod domain is indicated by double arrows, and the helical regions separated by linear linkers L1 and L2 (Fig. 1C) are indicated by single arrows. MESENCHYMAL PROGENITOR PROLIFERATION 359 Fig. 3. Amino acid comparison of partial amino acid sequences of NvK18 with Xenopus K18 (XK18), mouse K18 (MK18), human K16 (HK16), and human K17 (HK17). (A) Part of helix 1B and of the linker between helix 1 and 2. (B) Part of the C-terminus. Identical amino acids are in bold, and an amino acid motif at the end of the C-terminus (DGKVVS) that is similar to a motif found in the same position in other K18, in HK17, and in K8 in newt and other species is underlined. also in the epithelium of the limb blastema, their pattern of expression in the regenerating limb was examined by in situ hybridization. Figure 5A,B shows that by 10 days after amputation (early blastema stage), both NvK8 and NvK18 are expressed in the majority of cells in the blastema mesenchyme but not in the wound epidermis. The time course of expression of the NvK8 transcript parallels that of the protein (Table 1), with the only exception being at day 3 after amputation, when the NVK8 mRNA but not the NVK8 protein could be detected. At this stage, only a few blastemal cells have accumulated beneath the wound epidermis, and cell proliferation has not yet begun or is just starting (the onset of DNA synthesis has been reported to occur approximately 4 days after amputation; Hay and Fishman, 1961). By 5 days after amputation, however, both transcript and protein are clearly detectable. In normal limbs, expression of these keratins is observed in subepidermal glands (not shown), consistent with K8 and K18 immunoreactivity observed in newts and other species (Ferretti et al., 1989). Also, cultured cells with the ability to differentiate into myotubes express high levels of NvK8 (Fig. 5C) and NvK18 (not shown) transcripts. These data demonstrate that undifferentiated newt limb mesenchymal cells do indeed express simple epithelial keratins both in vivo and in vitro. observed. Expression of NvK8 and NvK18 was analysed by in situ hybridization 2–3 days after amputation, when a well-formed blastema was apparent, and compared with that in the contralateral unamputated limb (Fig. 5D). Expression of NvK8 and NvK18 was detected neither in unamputated limb buds at any stage of development (Fig. 5D,E) nor in regenerating forelimbs amputated between stages 32 (Fig. 5E) and 39. In contrast, upregulation of NvK8 and NvK18 following amputation was observed from stage 41 onward, when the forelimb has become richly innervated (Fekete and Brockes, 1987); the keratin-positive regenerating limb bud of a larva amputated at stage 45 is shown in Figure 5F. Expression of NvK8 and NvK18 in Amputated and Normal Limb Buds Previous immunocytochemical studies in P. waltl embryos suggested that NvK8 and 18 are not expressed in the developing limb bud. To confirm this observation in N. viridescens and establish whether these transcripts are upregulated following amputation in developing and regenerating limbs, limbs from stage 32–45 embryos were amputated. In animals left to regenerate for 2–3 weeks, regeneration of normal limbs was always Comparison of the Expression of NvK8 in Adult Denervated and Normal Blastemas Since induction of NvK8 and NvK18 expression correlates with the onset of innervation during development and because these keratins are first detected in the adult regeneration blastema at the onset of the nerve-dependent proliferative phase (3–5 days after amputation, Table 1), we investigated whether the presence of innervation was causally related to keratin expression in adult blastemas. Limbs denervated 2 days before amputation were analysed 7 days later (5-day regenerates) for the expression of NvK8 and NvK18 either by immunocytochemistry using the LP1K and RGE53 antibodies (Fig. 6) or by in situ hybridisation (not shown). We consistently found significant Nvk8 (Fig. 6A) and NvK18 (Fig. 6B) reactivity in blastemal cells of denervated limbs, suggesting that the presence of the nerve is not required for their induction. However, this experiment did not rule out the possibility that keratins have to be expressed for blastemal cells to maintain the undifferentiated state and respond to the nerve-derived mitotic signals. 360 CORCORAN AND FERRETTI Effect of NvK8 and NvK18 Anti-Sense Oligomers on DNA Synthesis and Phenotype of Cultured Limb Cells To address the issue of whether there is any correlation between keratin expression and the ability of blastemal cells to proliferate, we have taken advantage of two newt cell lines (Ferretti and Brockes, 1988) originated from normal limbs (TH4B) and from limb blastemas (BlH1) which express NvK8 and NvK18 (Fig. 5C). To downregulate keratin expression, these cells were treated for 7 days with anti-sense oligomers targeted against NvK8 and NvK18 (Fig. 1C), either separately or in combination. Scrambled oligomers were used as controls. This relatively long treatment time was chosen because of the long half-life of intermediate filament proteins. At the highest concentration of oligonucleotides used, 5 µM, either separately or in combination, some toxic effect was observed. Nonetheless, as shown in the example given in Figure 7A, when cells were treated with 2.5 µM each of K8 and K18 oligomers, DNA synthesis was significantly lower both in the BlH1 and TH4B cultures treated with anti-sense oligomers than in those treated with scrambled ones (Fig. 7A). This suggests that anti-sense oligomers to NvK8 and NvK18 have a specific effect on cell growth. When cells were treated with NvK18 anti-sense and scrambled oligomers at a lower concentration, 2.5 µM, 3H-thymidine incorporation was not affected by the scrambled oligomer but was significantly decreased in anti-sense– treated cells (Fig. 7B). A decrease in 3H-thymidine incorporation was also observed when cells were treated with 1 µM anti-sense oligomer, although the experimental variability was much higher, and when the same concentrations of NvK8 rather than NvK18 anti-sense were used (data not shown). To confirm that keratin anti-sense oligomer treatment could indeed lead to a decrease in keratin levels, we used an antibody specific for K18 (Ramaekers et al., 1984; Ferretti et al., 1989) in an enzyme-linked immunoadsorbent assay (ELISA) to measure changes in NvK18 Fig. 4. Tissue distribution of NvK8 and NvK18 transcripts. (a,b) Northern blot analysis. Lane 1, normal adult forelimb; lane 2, proximal forelimb blastema; lane 3, liver; lane 4, intestine; lane 5, esophagus with the 300 nucleotide PstI-PstI NvK8 probe (a) and the 290 nucleotide BglII-EcoRI NvK18 probe (b). Five µg poly(A)1RNA per lane were used. The size of the transcripts were estimated from the mobility of newt 28S and 18S ribosomal RNAs. Exposure times were 2 days with intensifying screens. Note that high levels of expression of both NvK8 and NvK18 are found in the regenerating blastema, liver, and intestine, but no expression is detected either in normal limb or esophagus. (c) RNase protection analysis of NvK8 expression in limbs and tails with the PstI-PstI probe. Normalised RNA samples (5 µg per lane) were used. Lane 1, input probe; lane 2, fingertip; lane 3, hand; lane 4, arm; lane 5, proximal forelimb blastema; lane 6, distal forelimb blastema; lane 7, tail blastema; lane 8, normal tail; lane 9, tail without tip; lane 10, tRNA. The size of the protected fragment is indicated. Note the significant increased in NvK8 expression both in regenerating limbs and tails. MESENCHYMAL PROGENITOR PROLIFERATION 361 Fig. 5. Localization of NvK8 and NvK18 transcripts by in situ hybridisation in adult limb blastemas (A,B), in cultured myogenic limb cells (C), and in normal and amputated developing limbs (D–F). (A) Expression of the NvK8 transcript in a distal 10-day blastema. (B) Expression of the NvK18 transcript in a proximal 10-day blastema. Note that NvK8 and NvK18 are expressed in most blastemal cells but not in the wound epidermis (WE in A and B). (C) Expression of NvK18 transcript in undifferentiated cultured limb cells. (D) Expression of NvK18 in the normal developing limb bud (dl) and contralateral regenerating limb bud (rl) of a stage 32 embryo. Both limb buds are negative; positive tissues include the area surrounding the pharyngeal cavity, liver, and gut. (E) Expression of the NvK8 transcript in the limb of a stage 41 larva. The developing limb is negative. High levels of the transcript are detectable in the notochord (n); some expression is also detectable in the gills, some of which are indicated (g). (F) Expression of the NvK18 transcript in a regenerating limb amputated at stage 45; the NvK18-positive regenerating limb bud is indicated by arrowheads. g, gills; e, eye. Scale bars 5 100 µm. A,B and D,E are at the same magnification. expression following 1 week treatment with 2.5 µM anti-sense. As shown above, this treatment induces a significant decrease in 3H-thymidine incorporation. Both in TH4B and BlH1 cells treated with NvK18 anti-sense, the levels of NvK18 protein were significantly lower than in controls (Fig. 8). These results confirm that the anti-sense oligomer downregulates keratin expression and suggest a causal relationship between the decrease in keratin expression and inhibition of DNA synthesis. Finally, the morphology of the cells was carefully monitored throughout the duration of each experiment to further confirm that the changes observed were not 362 CORCORAN AND FERRETTI TABLE 1. Expression of NvK8 in Limb Blastemas Proteina mRNAb 3 2 1 Days after amputation 5 7 111 111 111 111 14 111 111 aThe NvK8 protein was detected in cryostat sections by LP1K reactivity. bThe NvK8 mRNA was detected by in situ hybridization and at 14 days also by RNAase protection. in any way due to toxicity. We did not find any evidence of toxic effect in cultures treated with 2.5-µM oligomers, but we observed morphological changes in cells treated with either NvK18 or NvK8 anti-sense oligomers that were rather striking by 7 days of treatment (Fig. 9). In a typical experiment, TH4B cells in five of six wells and BlH1 cells in 4 of six wells had acquired a bipolar morphology, and some of them had fused (Fig. 9 A,B). Bipolar morphology in these newt cells reflects progression toward myogenic differentiation (Ferretti and Brockes, 1988), as also confirmed in the present experiments by the occurrence of myotube formation (Fig. 9 A,B). In control cultures maintained either in normal serum or treated with scrambled oligomers, cells with bipolar morphology were very rare (Fig. 9C–F). Therefore, there is a significant correlation between downregulation of NvK18 and NvK8 and changes in the cell phenotype. DISCUSSION Newt K8 and 18 are More Similar to Xenopus and Fish Than to Mammalian K8 and 18 Cloning and expression analysis of the newt homologues of K8 and K18 (NvK8 and NvK18, respectively) has firmly established that these simple epithelial keratins are expressed in the mesenchymal progenitor cells of the regenerating newt limb and in cultured blastemal cells with myogenic potential. Sequence comparison at the amino acid level of NvK8 with another type II newt keratin, NvKII, and with K8s from other species has shown that sequence conservation between NvK8 and NvKII is lower than the conservation between NvK8 and mammalian K8s. This is consistent with our current understanding of the evolutionary changes of keratin genes. The tail of NvK8 is longer than that of mammalian simple epithelia keratins and rich in glycine, like that of Xenopus and goldfish K8 (Franz and Franke, 1986; Giordano et al., 1989). In mammalian cytokeratins, glycine-rich repeats are only found in the tail of keratins typical of stratified epithelia (Hanukoglu and Fuchs, 1983; Jorcano et al., 1984). The presence of these glycine-rich motifs in amphibians and fish but not in mammalian K8 may reflect not only the likelihood that lower vertebrate K8 is an ancestral cytokeratin, but also different functional roles of this keratin in lower and higher vertebrates. In fact, the tail of intermediate filaments has been shown to have a role in filament stabilization and cellular localization (Wilson et al., 1992; Rogers et al., 1995). The partial sequence of NvK18 clearly shows that this keratin is the newt homologue of K18 proteins in other vertebrates and that this protein has been highly conserved through evolution. NvK8 and Nv18 Are Expressed in the Mesenchyme of Regenerating Adult Limbs but not of Regenerating Embryonic Limb Buds In situ hybridization analysis of NvK8 and NvK18 expression in regenerating and developing limbs has demonstrated that although these keratins are indeed expressed in the mesenchyme of regenerating adult limbs, the developing limb does not express them either at the early bud stage or at later stages of development. This is consistent with previous observations in the embryo of a different urodele species, P. waltl (Ferretti et al., 1989), and suggests that this difference in keratin expression in developing and regenerating limbs is a feature common to urodeles rather than a speciesspecific phenomenon. In addition, it indicates that co-expression of vimentin and keratin is not a normal attribute of limb mesenchyme but is associated with the regenerative process. The different phenotype of progenitor cells in developing and regenerating limbs may reflect the different origin of these cells in the two systems. Blastemal cells are believed to originate from dedifferentiation of the mature tissues of the stump, a process that would not be required when an early limb bud is amputated because it is still rather undifferentiated. Regrowth of the missing part could therefore be achieved through a regulative rather than regenerative process. In addition, growth of these progenitor cells is nerve-independent, unlike in adult or more developed limbs. Therefore, the transition between keratin nonexpressing and expressing cells in regenerating limbs could be related to the occurrence of dedifferentiation and nerve dependency, as discussed below. Expression of NvK8 and 18 may also reflect the fact that when dedifferentiated cells accumulate beneath the wound epidermis to form the blastema they are very loosely associated rather than tightly packed like the progenitor cells of developing limb buds. Therefore, blastemal cells might need a more rigid cytoskeleton to maintain their shape and position, and this could be accomplished by expressing both keratin and vimentin filaments rather than only vimentin. Finally, because cell motility and co-expression of the keratin pair 8-18 and vimentin have been found to be associated with invasiveness in certain tumors (Chu et al., 1993), it is conceivable that K8-18 expression in conjunction with vimentin in the loose blastemal mesenchyme may have a role in supporting cell movement and spatial organization during regeneration. These processes may require an additional level of control given the different ‘‘struc- MESENCHYMAL PROGENITOR PROLIFERATION 363 Fig. 6. Keratin expression (red) in 5-day limb regenerates denervated 2 days before amputation; nuclei are in blue. (A) RGE53 immunoreactivity in a denervated blastema; note that expression of NVK18 is clearly detectable in blastemal cells. (B) LP1K immunoreactivity in a denervated blastema; expression of NVK8 is clearly detectable in blastemal cells and in subepidermal glands (arrows). A few positive cells (arrowheads) are also observed in the thickened wound epidermis (WE), where LP1K reacts with another type II keratin, NvKII (Ferretti et al., 1991, 1993; Ferretti and Ghosh, 1997). Scale bar 5 100 µm. 364 CORCORAN AND FERRETTI Fig. 8. Changes in NvK18 protein in BlH1 and TH4B cells following 7-day treatment with 2.5 µM scrambled and anti-sense NvK18 oligomers was evaluated by ELISA immunoassay using the mAb CK18.2. The data are expressed as percentage of controls in normal medium. Both in BlH1 and TH4B cultures, the levels of the NvK18 protein are significantly lower in anti-sense–treated than in scrambled-treated cells or in normal medium (*P , 0.0001 by Student’s t-test; n 5 6; bars 5 SE). Keratin Expression in Regenerating Limbs Appears to be Associated With Innervation but Is not Controlled by it Fig. 7. Effect of 1-week treatment with NvK8 and NvK18 oligomers, either separately or in combination, on 3H-thymidine incorporation in BlH1 and TH4B cells. The data are expressed as percentage of controls in normal medium. (A) Treatment with 2.5 µM each of NvK8 and NvK18 scrambled and anti-sense oligomers. Some toxic effect is observed at this concentration (5 µM total) of oligomers; however, the effect of anti-sense oligomers in reducing 3H-thymidine incorporation is significantly higher in anti-sense than scrambled oligomer-treated cultures (*P , 0.001). (B) Treatment with 2.5 µM NvK18 scrambled and anti-sense oligomers. Note that at this concentration the scrambled oligomer does not affect 3Hthymidine incorporation, which in contrast is significantly reduced in anti-sense–treated BlH1 and TH4B cells (*P , 0.0001 by Student’s t-test; n 5 6; bars 5 SE). ture’’ of the limb mesenchyme in developing and regenerating limbs and the greater scale, both in term of volume and number of cells involved, of the regenerative process as compared to development (Ferretti and Brockes, 1991). In addition to comparing developing limb buds and adult limb blastemas, we have examined NvK8 and 18 expression in regenerating limbs that had been amputated at different developmental stages. We have shown that keratin expression is not induced in the regenerating mesenchyme when the developing limb is amputated before it becomes innervated, but is induced in more mature, innervated limbs like in the adult. A similar response to amputation in developing and adult limbs has been previously reported in the case of the 22/18 antigen. This phenotypic change has been taken to reflect a transition controlled by the nerve from nerve-independent to nerve-dependent growth of blastemal cells (Fekete and Brockes, 1987, 1988). Our studies have identified two additional molecules, the developmentally regulated and regeneration-associated simple epithelia keratins 8 and 18, which reflect this transition. Notwithstanding these similarities between 22/18 and keratin expression, there is a significant difference in their time course of expression in the adult regeneration blastema and in the type of regulation. Whereas 22/18 is detected within 24 hr after amputation (Gordon and Brockes, 1988), and this is most likely due to a conformational change of the protein rather than to de novo synthesis (Ferretti and Brockes, 1990), keratin transcripts are first detected at 3 days and the proteins slightly later, when the blastemal cells start to divide MESENCHYMAL PROGENITOR PROLIFERATION 365 Fig. 9. Effect of 2.5 µM anti-sense, either NvK8 or NvK18, oligomers on the phenotype of BlH1 (A,C,E) and TH4B (B,D,F) cells. (A) NvK8 anti-sense–treated BlH1; note the presence in this culture of bipolar cells, some of which have fused or are fusing (arrows). (B) NvK18 anti-sense– treated TH4B; note the presence in this culture of bipolar cells, some of which have fused or are fusing (arrows). (C) Untreated BlH1; note the absence of bipolar cells. (D) Untreated TH4B; note the absence of bipolar cells. (E) NvK8 scrambled-treated BlH1; note that these cells are not significantly different from untreated controls in C. (F) NvK8 scrambledtreated TH4B; note that these cells are not significantly different from untreated controls in D. Scale bar 5 50 µm. under the influence of the nerve (Fig. 10A). Although the pattern of expression of K8 and 18 during development and regeneration suggests that the nerve itself might be required to control their expression, our experimental results have clearly shown that this is not the case. Although we cannot completely rule out a decrease in K8 or 18 levels in denervated animals because of the nonquantitative techniques used in these experiments (immunocytochemistry and in situ hybridization), both proteins are still clearly detectable in 5-day denervated blastemas. This is consistent with the fact that the initial accumulation of blastemal cells is not nerve-dependent and does not rule out the possibility that keratin expression is a prerequisite for limb blastemal cells to start to divide. This possibility is consistent not only with their pattern of expression in embryonic and adult regenerating limbs, but also with the observation that even in denervated limbs, a fraction of proliferating cells is present 4 days after amputation (Barger and Tassava, 1985). Downregulation of NvK8 and NvK18 in the Newt Limb Cells Is Associated With Inhibition of Proliferation and Differentiation To test the hypothesis of a causal relationship between keratin expression and the ability of blastemal cells to divide, we have used anti-sense NVK8 or NvK18 oligonucleotides to downregulate these keratins in cultured newt cells. Because NVK8 or NvK18 are obligatory heteropolymers, downregulation of either of them 366 CORCORAN AND FERRETTI Fig. 10. Summary. (A) The onset of keratin expression in vivo following amputation coincides with the beginning of nerve-dependent proliferation of blastemal cells and is downregulated when cells stop dividing and differentiate into muscle. (B) Downregulation of keratin expression in vitro coincides with a significant inhibition of proliferation and the onset of myogenic differentiation. would impair filament formation, and downregulation of either protein could be expected to produce the same effects. This was indeed the case, and we have found a marked decrease in cell proliferation in cultures treated either with anti-sense NVK8 or NvK18 oligonucleotides or with the two in combination, which is significantly higher than that of scrambled oligonucleotides even at a concentration at which some toxic effects occur (5 µM). The fact that we do not see complete mitotic arrest, as in serum-deprived cultures (not shown), may be due to the fact that treatment with anti-sense oligomers at nontoxic concentrations does not result in the complete disappearance of the keratin proteins. Nonetheless, the significant decrease in protein levels and changes in the cell phenotype (the switch to a bipolar morphology in these cells is indicative of the onset of myogenic differentiation; Ferretti and Brockes, 1988) in the NvK18 anti-sense–treated cultures, as compared with controls, supports the view that there is a causal relationship between keratin expression, proliferation, and differentiation. This is not entirely surprising because it has recently become apparent that intermediate filaments and their associated proteins are highly dynamic structures; it has also been proposed that in addition to their structural functions as integrators of the intracellular space, they may be active players at different stages of the cell cycle (Foisner, 1997). However, it is presently difficult to assess whether keratins can somehow play a direct role in the control of proliferation, and it is the inhibition of proliferation following keratin downregulation which induces differentiation, or whether keratin downregulation induces differentiation and, as a consequence, the cells exit the mitotic cycle (see Fig. 10B). Notwithstanding suggestions that intermediate filaments can act as transcription factors (Traub and Shoeman, 1994), the possibility that this may be the way they control division in blastemal cells seems see rather remote. What may be more likely, although at this stage highly speculative, is the possibility that keratins are somehow involved in the mitotic cycle indirectly, through their ability to bind/sequester factors important in the control of the cell cycle. For example, the fact that intermediate filaments bind plectin, which is thought to integrate the organization of actin and intermediate filaments in a cell-cycle and phosphorylation-dependent fashion, has been recently demonstrated (Foisner et al., 1995), as has the binding between K8/18 filaments and proteins of the 14-3-3 family, which interact with several signal transduction kinases (Liao and Omary, 1996). It has also been shown that the Retinoblastoma (Rb) protein in the hyperphosphorylated state binds to lamins, and it has been suggested that disassembly of such complex following phosphorylation events may be required for cell cycle progression (Mancini et al., 1994). Finally, intermediate filaments connect the nuclear and cell membrane, and it has been shown that expression of keratins can be modulated by the extracellular matrix and that their interaction with it is associated with changes in the motile behaviour of certain cell types (Chu et al., 1993; Hendrix et al., 1997). This suggests that these molecules can ‘‘read,’’ either directly or indirectly, molecular cues presumably mediated by membrane receptors. Therefore, it is not inconceivable that a decrease in keratin filaments in blastemal cells may affect intracellular communication between the plasma membrane and the nucleus, possibly by affecting the function of keratin-associated molecules involved in growth control, and that consequently keratin expression is required to make blastemal cells responsive to the nerve. As mentioned above, another possible explanation for our results is that keratins play a role in the maintenance of the undifferentiated state, rather than in proliferation, and that their downregulation may favour myogenic differentiation. It has been shown that other members of the intermediate filament family, such as desmin, can play a role in differentiation in vitro (Li et al., 1994; Bahler, 1996). In addition, overexpression of the glial intermediate filament GFAP in an astrocytoma cell line has been shown to induce changes in cell shape and inhibit growth (Toda et al., 1994). How intermediate filaments might control differentiation is rather unclear, but arguments similar to those discussed above could be invoked to explain loss of the MESENCHYMAL PROGENITOR PROLIFERATION dedifferentiated state/induction of differentiation. In addition, it has been shown that K18 transcription can be stimulated through its Ets-1 and AP-1 responsive elements by the Ras signal transduction pathways, suggesting that its expression may be important in the behaviour of certain tumors (Pankov et al., 1994). For example, expression of both K8 and 18 in conjunction with vimentin has been associated with tumor invasiveness (Chu et al., 1993; Hendrix et al., 1997), and this suggests that K8 and 18 are not necessarily terminal targets of the cascade of events controlling differentiation, but are somewhere more upstream and might exercise some ‘‘control’’ over this pathway themselves. A parallel between certain tumor and blastemal cells is that they are undifferentiated/poorly differentiated, highly proliferative, motile cells found in adult organisms that express a mixed (vimentin and keratin) phenotype; the main difference is that the growth of blastemal cells, unlike that of tumors, is under strict control and they eventually differentiate. In conclusion, our results indicate a casual relationship between NvK8-18 expression in limb blastemal cells, the undifferentiated state and proliferation, since cells stop dividing and undergo myogenic differentiation when these keratins are downregulated. However, it is not currently possible to say whether the primary effect of keratin downregulation is on proliferation or differentiation, and it will be a significant challenge to answer this complex question. EXPERIMENTAL PROCEDURES Animals and Surgery Adult N. viridescens supplied by Sullivan & Co. (Nashville, TN) were maintained at 20°C and fed shredded bovine heart on alternate days. All surgical procedures were performed on newts anaesthetized in 0.1% tricaine (3-aminobenzoic acid ethylester methanesulphate salt; Sigma, St. Louis, MO). Following bilateral forelimb amputation at either the level of the mid-humerus (proximal amputation) or the mid-radius/ ulna (distal amputation) and recovery in 0.5% sulfamerazine (Sigma), the regenerating newts were maintained at 25°C. Denervated blastemas were obtained by severing the right limb brachial nerves at the midscapular level 2 days before proximal limb amputation. Regeneration blastemas were harvested at different times following amputation (3, 5, 10, 15 days, depending on the experiment to be performed). N. viridescens embryos were obtained by injecting pregnant females with human chorionic gonadotropin (HCG; Sigma) as previously described (Ghosh et al., 1994). Amputation of developing limbs was performed unilaterally in embryos between stage 32 (early limb bud) and 45 (3 digit stage). Embryos were killed by an overdose of tricaine and processed for in situ hybridization 2–3 days after amputation, when a regeneration blastema was always apparent. To check that regeneration could ensue normally, some animals were left to regenerate for 2–3 weeks. 367 Isolation and Analysis of cDNA Clones A cDNA library in lZapRII from cultured blastemal cells (gift of J.P. Brockes) was screened with the mAb LP1K to isolate cDNAs encoding the cytoskeletal protein of apparent molecular weight 52 kD recognized by this mAb in blastemal cells. Five 3 105 pfu of the amplified library were plated onto Escherichia coli strain Y1090rk-; after 3–4 hr incubation at 37°C, nitrocellulose filters (previously soaked in 10 mM IPTG and dried) were laid over the plates. The filters were processed and reacted with LP1K as previously described (Ferretti et al., 1991). Selected LP1K-positive plaques were purified, Bluescript SK- plasmids containing the inserts of interest were excised and propagated in the E. coli strain XL1blue, and the cDNA clones were analyzed for their size, restriction map, and nucleotide composition. The same cDNA library was also screened for the newt homologue of K18 with a human K18 DNA probe (USB, Cleveland, OH). The dideoxy sequencing reactions were performed using the Sequenase DNA sequencing kit (United States Biochemical). Sequences were compared using the software from the University of Wisconsin Genetics Computer Group (UWGCG) and the ‘‘Gap’’ programme used for alignment of amino acid sequences. RNA Preparation and Analysis Blastemas, cultured cells, and other tissues to be used for RNA analysis were harvested, pooled, and stored in liquid nitrogen until the extraction was performed. RNA was extracted from newt tissues by the guanidine isothiocyanate procedure as previously described (Brown and Brockes, 1991). A 300 bp PstI-PstI fragment, which encodes part of the a-helix 1B of NvK8 (Fig. 1A), and a 290 bp BglII-EcoRI fragment, which encodes part of the a-helix 1A of NvK18 (Fig. 1B), were used to prepare DNA probes for Northern blotting, and riboprobes for RNAase protection and in situ hybridization experiments. Either 10 µg of total RNA or 5 µg of poly(A)1RNA isolated by using the polyATract mRNA system IV (Promega, Southampton, UK) were used for Northern blot analysis, which was performed essentially as previously described (Casimir et al., 1988). Hybridization was carried out in 50% formamide at 42°C, and the filters were washed twice for 15 min at room temperature and once for 30 min at 42°C with 0.15 M SSC containing 0.1% SDS. The labelled riboprobe for RNAase protection analysis was hybridized with 5 µg of total RNA, and the assay was essentially performed according to Casimir et al. (1988). All the RNA samples were normalized by using the satellite 2 DNA probe pSP6D6 (Epstein and Gall, 1987) in conjunction with measurement of the O.D. at 260 nm. In situ hybridization was performed as previously described (Ghosh et al., 1996). In brief, tissues were fixed overnight at 4°C in 4% paraformaldehyde (PFA) in A-PBS, paraffin wax-embedded, and 6-µm sections cut. 368 CORCORAN AND FERRETTI Dewaxed sections were fixed in 4% PFA for 20 min, treated with 20 µg/ml of proteinase K, post-fixed, acetylated, and hybridized overnight at 55°C with 105 cpm/ml of either the transcribed NvK8 (PstI-PstI) or NvK18 (BglII-EcoRI) riboprobe (Fig. 1 A,B). Controls with sense experiments and with another keratin probe, NvKII (Ferretti et al., 1991; Ferretti and Ghosh, 1997), were also carried out and confirmed the specificity of the signal detected with the anti-sense NvK8 and NvK18 probes. After high-stringency washing (65°C with 50% formamide, 2 3 SSC, 10 mM DDT) and RNAase A digestion for 30 min at 37°C followed by further washing as previously described (Ghosh et al., 1996), the slides were dehydrated and processed for autoradiography. Slides were and exposed for 5 days at 4°C before being developed and counterstained with toluidine blue. effect of treatment with scrambled and anti-sense oligomers was compared by Student’s t-test. Proliferation Assay Immunohistochemistry was performed as previously described using LP1K to detect NvK8 and either RGE53 or CK18.2 to detect NvK18 (Ramaekers et al., 1984; Broers et al., 1986). The bound antibody was detected by the nuclei stained with Hoechst dye (Ferretti et al., 1989). The effect of 2.5 µM anti-sense and scrambled oligomers on NvK18 and NvK8 protein levels was evaluated by ELISA using CK18.2 and LP1K supernatants diluted 1:2. As in the thymidine incorporation assay, cells were plated at an initial density of 750 cells/well and grown for 1 week in the presence of different oligomers. At the end of the treatment, the tissue culture medium was removed and the wells washed with 70% L-15. The cells were fixed with cold acid-alcohol for 10 min. The fixed cells were processed as for immunocytochemistry, except that the secondary antibody used was an affinity purified alkaline-phosphatase-conjugated rabbit antiserum to mouse immunoglobulins (Dako) diluted in newt culture medium. After 45 min incubation with the secondary antibody, the wells were rinsed three times with PBS-0.05% Tween-20 and then with 0.1 M Tris, pH 9.5. Two hundred µl of r-Nitrophenyl Phosphate substrate solution (Sigma FAST rNPP) was then added to each well, and the plate read at 405 nm using an automatic plate reader after a 30-min incubation. Each experiment was repeated at least twice, and each group consisted of at least five samples. The effect of treatment with scrambled and anti-sense oligomers was compared by Student’s t-test. The effect of anti-sense oligomers on DNA incorporation was studied in long-term cultures of limb cells that express the blastemal phenotype. These long-term cultures originated from cells that had migrated from small explants of either normal limbs (TH4B) or limb blastemas (BlH1). Cells were grown and passaged essentially as previously described (Ferretti and Brockes, 1988), apart from the fact that 0.1% bovine skin gelatin (Sigma) instead of collagen was used to coat the dishes. Eighteen base anti-sense oligomers targeted against NvK8 and NvK18 modified at both ends with phosphorothioate to increase their stability and purified by reverse-phase high-pressure liquid chromatography (GENOSYS, Cambridge, UK) were used at 1, 2.5, and 5 µM. The relative position and sequence of the two anti-sense oligomers used is shown in Figure 1C. The NvK8 anti-sense oligomer is targeted against part of helix 2 which has been shown to deplete Xenopus K8 (Heasman et al., 1992). The NvK18 anti-sense oligomer is targeted against the border between helix 1B and the linker between helix 1 and 2 (L12). To control for nonspecific effects, such as toxicity, scrambled sequences of the NvK8 and NvK18 anti-sense were used as controls. For DNA incorporation assays, 750 cells/well were plated in 96-well plates (Gibco, Paisley, UK). The treatment was started the day after plating, when the cells had attached and spread onto the substrate. Fresh oligomers were added every 2 days; on the sixth day from the beginning of the treatment, 3H-thymidine (methyl-18, 28-3H-Thymidine, 124 Ci/mmol, Amersham, Slough, UK) was added to medium at 1 mCi/ml. Cells were collected 24 hr later on filter using an automatic cell harvester. The filters were washed and dried, and the incorporated radioactivity measured by liquid scintillation counting. Experiments were repeated at least twice, and more usually three times; each experimental group always consisted of at least five samples. The Immunohistochemistry Immunohistochemistry was essentially performed as previously described (Ferretti et al., 1989). Reactivity of the anti-keratin mAbs LP1K (Lane et al., 1985), RGE53, and CK18.2 (Ramaekers et al., 1984; Broers et al., 1986) was assayed on 8-µm cryostat sections briefly fixed with cold acid-alcohol (95% ethanol–5% acetic acid). Bound antibodies were detected by a rhodamineconjugated rabbit anti-mouse immunoglobulin antibody (Rh-Rb-aMIg, Dako, High Wycombe, UK), and the nuclei stained with 1.25 µg/ml of Hoechst dye 33258 (Sigma). The results were recorded using an Axiophot (Zeiss) fluorescent photomicroscope. Protein Analysis Morphological Analysis The morphology of the cells was monitored throughout the duration of each experiment, and micrographs of live cells in 96-well plates were taken using an IM45 inverted microscope (Zeiss). ACKNOWLEDGMENTS The authors thank E.B. Lane and F. Ramaekers for the gifts of LP1K and CK18.1, respectively; J.P. Brockes for the gift of the cultured blastema cDNA library; R. MESENCHYMAL PROGENITOR PROLIFERATION Oshima for the gift of human K18 DNA probe; and P. Thorogood for reading the manuscript. This work was supported by a Wellcome Trust grant to P.F. REFERENCES Bahler M. Myosins on the move to signal transduction. Curr. Opin. Cell Biol. 1996;8:18–22. Barger PM, Tassava RA. Kinetics of cell cycle entry in innervated and denervated forelimb stumps of larval Ambystoma. J. Exp. Zool. 1985;233:151–154. Baribault H, Price J, Miyai K, Oshima RG. Mid-gestational lethality in mice lacking keratin 8. Genes Dev. 1993;7:1191–1202. Broers JL, Carney DN, Klein Rot M, Schaart G, Loue EB, Vooijs P, Ramaekers F. Intermediate filaments proteins in classic and variant types of small lung carcinoma cell line: A biochemical and immunological analysis using a panel of monoclonal and polyclonal antibodies. J. Cell Sci. 1986;83:37–60. Brown R, Brockes JP. Identification and expression of a regenerationspecific homeobox gene in the newt limb blastema. Development 1991;111:489–496. Casimir CM, Gates PB, Ross-Macdonald PB, Jackson JF, Patient RK, Brockes JP. Structure and expression of a newt cardio-skeletal myosin gene. Implications for the C value paradox. J. Mol. Biol. 1988;202:287–296. Charlebois TS, Spencer DH, Tarkington SK, Henry JJ, Grainger R. M. Isolation of a chick cytokeratin cDNA clone indicative of regional specialization in early embryonic ectoderm. Development 1990;108: 33–45. Chu YW, Runyan RB, Oshima RG, Hendrix MJC. Expression of complete keratin filaments in mouse L cells augments cell migration and invasion. Proc. Natl. Acad. Sci. USA 1993;90:4261–4265. Epstein LM, Gall JG. Self-cleaving transcripts of satellite DNA from the newt. Cell 1987;48:535–543. Fekete DM, Brockes JP. A monoclonal antibody detects a difference in the cellular composition of developing and regenerating limbs of newts. Development 1987;99:589–602. Fekete DM, Brockes JP. Evidence that the nerve controls molecular identity of progenitor cells for limb regeneration. Development 1988;103:567–573. Ferretti P, Brockes JP. Culture of newt cells from different tissues and their expression of a regeneration-associated antigen. J. Exp. Zool. 1988;247:77–91. Ferretti P, Brockes JP. The monoclonal antibody 22/18 recognizes a conformational change in an intermediate filament of the newt, Notophthalmus viridescens, during limb regeneration. Cell Tissue Res. 1990;259:483–493. Ferretti P, Brockes JP. Cell origin and identity in limb regeneration and development. Glia 1991;4:214–224. Ferretti P, Ghosh S. Expression of regeneration-associated cytoskeletal proteins reveals differences and similarities regenerating organs. Dev. Dyn. 1997;210:288–304. Ferretti P, Fekete DM, Patterson M, Lane EB. Transient expression of simple epithelial keratins by mesenchymal cells of regenerating newt limb. Dev. Biol. 1989;133:415–424. Ferretti P, Brockes JP, Brown R. A newt type II keratin restricted to normal and regenerating limbs and tails is responsive to retinoic acid. Development 1991;111:497–507. Ferretti P, Corcoran JP, Ghosh S. Expression and regulation of keratins in the wound epithelium and mesenchyme of the regenerating newt limb. Prog. Clin. Biol. Res. 1993;383A:261–269. Foisner R. Dynamic organisation of intermediate filaments and associated proteins during the cell cycle. BioEssays 1997;19:297–305. Foisner R, Bohn W, Mannweiler K, Wiche G. Distribution and ultrastructure of plectin arrays in subclones of rat glioma c-6 cells differing in intermediate filament protein (vimentin) expression. J. Struct. Biol. 1995;115:304–317. Franz JK, Franke WW. Cloning of cDNA of a cytokeratin expressed in oocytes of Xenopus laevis. Proc. Natl. Acad. Sci. USA 1986;83:6475– 6479. Géraudie J, Ferretti P. Gene expression during amphibian limb regeneration. Int. Rev. Cytol. 1997;in press. 369 Ghosh S, Thorogood P, Ferretti P. Regenerative capability of upper and lower jaws in urodele amphibians. Int. J. Dev. Biol. 1994;38:479– 490. Ghosh S, Thorogood P, Ferretti P. Regeneration of lower and upper jaws in urodeles is differentially affected by retinoic acid. Int. J. Dev. Biol. 1996;40:1161–1170. Giordano S, Glasgow E, Tesser P, Schechter N. A type II keratin is expressed in glial cells of the goldfish visual system. Neuron 1989;2:1507–1516. Gordon H, Brockes JP. Appearance and regulation of an antigen associated with limb regeneration in Notophthalmus viridescens. J. Exp. Zool. 1988;247:232–243. Hay ED, Fishman DA. Origin of the blastema in regenerating limbs of the newt Triturus viridescens. An autoradiographic study using tritiated thymidine to follow cell proliferation and migration. Dev. Biol. 1961;3:49–77. Hanukoglu I, Fuchs E. The cDNA sequence of a type II cytoskeletal keratin reveals constant and variable structural domains among keratins. Cell 1983;33:915–924. Heasman J, Torpey N, Wylie C. The role of intermediate filaments in early Xenopus development studied by antisense depletion of maternal mRNA. Development Suppl. 1992;119–125. Hendrix MJC, Seftor EA, Seftor REB, Trevor KT. Experimental co-expression of vimentin and keratin intermediate filaments in human breast cancer cells results in phenotypic interconversion and increased invasive behavior. Am. J. Pathol. 1997;150:483–495. Jorcano JL, Franz JK, Franke WW. Amino acid sequence diversity between bovine epidermal cytokeratin polypeptides of the basic (type II) subfamily as determined from cDNA clones. Differentiation 1984;28:155–163. Ku NO, Michie S, Oshima RG, Omary MB. Chronic hepatitis, hepatocyte fragility, and increased soluble phosphoglycokeratins in transgenic mice expressing a keratin18 conserved arginine mutant. J. Cell Biol. 1995;3:1303–1314. Ku NO, Wright TL, Terrault NA, Gish R, Omary MB. Mutation of human keratin 18 in association with cryptogenic cirrhosis. J. Clin. Invest. 1997;99:19–23. LaFlamme SE, Jamrich M, Richter K, Sargent TD, Dawid IB. Xenopus endo B is a keratin preferentially expressed in the embryonic notochord. Genes Dev. 1988;2:853–862. Lane EB, Bartek J, Purkis PE, Leigh I. Keratin antigens in differentiating skin. Ann. NY Acad. Sci. 1985;445:241–258. Li H, Choudhary SK, Milner DJ, Munir MI, Kuisk IR, Capetanaki Y. Inhibition of desmin expression blocks myoblast fusion and interferes with the myogenic regulators myoD and myogenin. J. Cell Biol. 1994;124:827–841. Liao J, Omary MB. 14-3-3-protein associate with phosphorylated simple epithelial keratins during cell cycle progression and act as a solubility cofactor. J. Cell Biol. 1996;133:345–357. Mancini MA, Shan B, Nickerson JA, Penman S, Lee WH. The retinoblastoma gene product is a cell cycle-dependent, nuclear matrix-associated protein. Proc. Natl. Acad. Sci. USA 1994;91:418– 422. Oshima RG, Baribault H, Caulin C. Oncogenic regulation and function of keratins 8 and 18. Cancer Metastasis Rev. 1996;15:445–471. Pankov R, Umezawa A, Maki R, Der CJ, Hauser CA, Oshima RG. Oncogene activation of human keratin 18 transcription via the ras signal transduction pathway. Proc. Natl Acad. Sci. USA 1994;91:873– 877. Quax W, Egberts WV, Hendriks W, Quax Jeuken Y, Bloemendal H. The structure of the vimentin gene. Cell 1983;35:215–223. Ramaekers F, Schaap H, Mulder M, Huysmans A, Vooijs P. Cytokeratin filament expression during in vitro teratocarcinoma cell differentiation as detected by a monoclonal antibody. Cell Biol. Int. Rep. 1984;8:721–730. RayChaudhury A, Marchuk D, Lindhurst M, Fuchs E. Three tightly linked genes encoding human type I keratins: Conservation of sequence in the 58-untranslated leader and 58-upstream regions of coexpressed keratin genes. Mol. Cell Biol. 1986;6:539–548. Rogers KR, Eckelt A, Nimmrich V, Janssen KP, Schliwa M, Herrmann H, Franke WW. Truncation mutagenesis of the non-alpha-helical 370 CORCORAN AND FERRETTI carboxyterminal tail domain of vimentin reveals contributions to cellular localization but not to filament assembly. Eur. J. Cell Biol. 1995;66:136–150. Rosenberg M, RayChaudhury A, Shows TB, Le Beau MM, Fuchs E. A group of type I keratin genes on human chromosome 17: Characterization and expression. Mol. Cell Biol. 1988;8:722–736. Schotté OE, Butler EG. Phases of regeneration in the urodele limb and their dependence upon the nervous system. J. Exp. Zool. 1944;97:95– 121. Singer M. The growth and morphogenesis of the regenerating forelimb of adult Triturus following denervation at various stages of development. J. Exp. Zool. 1948;108:279–308. Singer M. The influence of nerve regeneration of the amphibian extremity. Quart. Rev. Biol. 1952;27:169–200. Singer PA, Trevor K, Oshima RG. Molecular cloning and characterization of the Endo B cytokeratin expressed in preimplantation mouse embryos. J. Biol. Chem. 1986;261:538–547. Thornton CS. Amphibian limb regeneration. Adv. Morphogen. 1968;7: 205–249. Toda M, Miura M, Asou H, Toya S, Uyemura K. Cell growth suppression of astrocytoma C6 cells by glial fibrillary acidic protein cDNA transfection. J. Neurochem. 1994;63:1975–1978. Traub P, Shoeman RL. Intermediate filament proteins: Cytoskeletal elements with gene regulatory function. Int. Rev. Cytol. 1994;154:1– 103. Wilson AK, Coulombe PA, Fuchs E. The roles of k5 and k14 head, tail, and R/KLLEGE domains in keratin filament assembly in vitro. J. Cell Biol. 1992;119:401–414.