An Extracellular Matrix Molecule of Newt and Axoloti Regenerating Limb Blastemas and Embryonic Limb Buds: Immunological Relationship of MT1 Antigen with Tenascin

Development 108, 657-668 (1990) 657 Printed in Great Britain ©The Company of Biologists Limited 1990

An extracellular matrix molecule of newt and axoloti regenerating limb blastemas and embryonic limb buds: immunological relationship of MT1 antigen with tenascin

HIROAKI ONDA, DAVID J. GOLDHAMER* and ROY A. TASSAVAt Department of Molecular Genetics, The Ohio State University, 484 W. 12th Ave., Columbus, Ohio 43210, USA 1 Author to whom correspondence should be addressed. 2 Present address: Department of Biology, University of Virginia, Charlottesville, Va. 22901, USA.

Summary

Several well-characterized extracellular matrix (ECM) period of embryonic limb development, mAb MT1 components have been localized to the amphibian limb reactivity was seen in the ECM of the mesenchyme and regenerate, but the identification and characterization of in a layer beneath the limb bud ectoderm, similar to its novel ECM molecules have received little attention. Here distribution during regeneration. Considerable mAb we describe, using mAb MT1 and immunocytochem- MT1 reactivity was also associated with the developing istry, an ECM molecule expressed during limb regener- somites. The reactivity of mAb MT1 in blastema and ation and limb development. In limb stumps, mAb MT1 limb bud was similar if not identical to that of a reactivity was restricted to tendons, myotendinous junc- polyclonal Ab against tenascin (pAbTN), a large, extra- tions, granules in the basal layers of epidermis, perios- cellular matrix glycoprotein implicated in growth con- teum (newts) and perichondrium (axolotls). In regener- trol, inductive interactions, and other developmental ating limbs, reactivity in the distal limb stump was first events. This pAbTN effectively competed against mAb detected 5 days and 1 day after amputation of newt and MT1 binding on blastema sections. In immunoblots, axoloti limbs, respectively. In both species, mAb MT1 both mAb MT1 and pAbTN recognized a very high 3 recognized what appeared to be an abundant blastema molecular weight (approximately MT1000 x 10 ) protein matrix antigen, localized in both thin and thick cords in blastema extracts of both newts and axolotls. mAb between and sometimes closely associated with blastema MT1 immunoprecipitated a protein of MT 1000K size cells. Reactivity was generally uniform throughout the which reacted to both mAb MT1 and pAbTN in immu- blastema except for a particularly thick layer that was noblots. These data show that tenascin is in the matrix of present immediately beneath the wound epithelium. the urodele blastema and limb bud, and suggest that During redifferentiation stages, mAb MT1 reactivity mAb MT1 identifies urodele tenascin. persisted among blastema cells and redifferentiating cartilage but was lost proximally in areas of muscle and Key words: limb regeneration, tenascin, extracellular connective tissue differentiation. During the entire matrix, monoclonal antibody.

Introduction 1982; Gulati et al. 1983) and neural cell adhesion molecule (N-CAM; Maier et al. 1986) have all been The interaction between embryonic cells and extra- shown to be developmental^ regulated during regener- cellular matrix (ECM) is emerging as a critical para- ation. Little is known of the role of these molecules in meter involved in dictating cell fates, proliferative regeneration, although hyaluronic acid synthesis has behavior, morphogenetic potential, and other develop- been implicated to be directly or indirectly dependent mental events (Hynes, 1981; Toole, 1981; Chiquet- upon the neurotrophic action of nerves (Smith et al. Ehrismann et al. 1986). To better define the roles of the 1975; Mescher and Munaim, 1986), and based on ECM during amphibian limb regeneration, the distri- antibody inhibition studies, Maier et al. (1986) have bution and timing of appearance of several major suggested an involvement of N-CAM in blastema ECM/cell surface components have been investigated. growth. Further studies of these and other matrix Collagen (Grillo etal. 1968), hyaluronicacid (Toole and components should reveal greater insights into their Gross, 1971), fibronectin and laminin (Repesh et al. involvement in the major reparative and developmental 658 H. Onda, D. J. Goldhamer and R. A. Tassava events which characterize limb regeneration, including its production and specificity have been reported previously wound healing, dedifferentiation, cell proliferation and (Chiquet-Ehrismann et al. 1986). This antiserum has been morphogenesis. shown to react to tenascin from different species, including During a search for molecules of developmental mammals and amphibians (Aufderheide et al. 1987; Epperlein et al 1988; Riou et al. 1988). The rabbit polyclonal antiserum significance to regeneration, we obtained a monoclonal (IgG fraction) against axolotl plasma fibronectin (pAbFN) antibody (mAb MT1; formerly referred to as mAb was a kind gift from Dr Y. Thouveny and Dr J. C. Boucaut, 4G3; Goldhamer and Tassava, 1986; Tassava, 1988) and its production and specificity have been described pre- reactive to the extracellular matrix of the blastema. viously (Boucaut and Darribere, 1983). This antiserum does Based on immunohistochemical observations, the dis- not react to tenascin, laminin, or entactin (Thouveny and tribution of the MT1 antigen in unamputated and Boucaut, personal communication), but reacts to both plasma regenerating limbs did not correspond with previously and cell surface fibronectin of amphibians (Boucaut and described molecules such as fibronectin, laminin, hyalu- Darribere, 1983). ronic acid, collagens, or N-CAM (Smith et al. 1975; Gulati et al. 1983; Maier et al. 1986), thus raising the Immunocytochemistry possibility that mAb MT1 recognizes a novel ECM Immunofluorescence methods were similar to those described molecule (Goldhamer and Tassava, 1986; Tassava, (Tassava et al. 1986; Goldhamer et al. 1989). In brief, freshly 1988). dissected unamputated limbs, regenerating limbs and embryos were frozen in OCT compound (Miles) in a slurry of dry We report here experiments designed to determine ice and isopropyl alcohol. Tissues were sectioned at 10/m\ on the identity of this matrix component and its temporal a cryostat and sections were allowed to dry at room tempera- and spatial distribution in regeneration and embryonic ture for 2h or overnight. Sections were used immediately or development. The results show that MT1 is an early stored at — 20°C. mAb MT1 was applied to sections as a lOx appearing antigen which is abundant in the blastema dilution of the stock solution in PBS. After 1 h of incubation, and limb bud matrix, but is restricted mainly to tendons slides were washed in PBS first with 0.05% Triton X-100 (three washes of 3min each) and then without Triton (one and skeletal element sheaths in mature limb tissues. wash of 3min). Secondary Ab (rhodamine labeled goat anti- Comparison of mAb MT1 reactivity with that of poly- mouse IgM; Cappel) was then applied for 1 h, after which clonal antibody to tenascin demonstrated a striking slides were washed as above and coverslipped. Controls were degree of similarity in regenerating and developing included as previously described (Tassava et al. 1986). limbs. Results from a series of tests, including immuno- For immunofluorescence staining with pAbTN and blotting and immunoprecipitation, are consistent with pAbFN, sections were first blocked with 5 % non-fat dry milk the view that mAb MT1 recognizes urodele tenascin. in PBS for 30 min, then incubated in diluted pAbTN (1:10), or pAbFN (1:10) for lh. The following steps were identical to those for mAb MT1 except that the secondary Ab was Materials and methods affinity-purified rhodamine-labeled goat anti-rabbit IgG (Cappel). General Newt and axolotl embryos were sampled for mAb MT1 Adult newts {Notophthalmus viridescens) were collected from reactivity at all stages of forelimb development; axolotl larvae ponds in southern Ohio. Axolotl (Ambystoma mexicanum) were additionally sampled at various stages of hindlimb larvae and embryos were obtained from the Indiana Univer- development. Both frontal and transverse cryostat sections, sity axolotl colony. Care, feeding, and surgical operations for including limb buds and various other body organs and both species have been described (Kelly and Tassava, 1973; tissues, were examined for reactivity to mAb MT1. Sections Mescher and Tassava, 1975). Axolotl larvae of 35, 45, and through limb buds were also examined for pAbTN reactivity. 75 mm snout-tail tip length were utilized. All operations were Reactivity to mAb MT1 was examined in distal regions of performed while animals were anesthetized with neutralized amputated axolotl limbs at 12 h, 24 h, and 4 days after MS 222 (ethyl m-aminobenzoate methanesulfonate; Sigma). amputation, and at 2 day intervals thereafter through digit Newt embryos were obtained as previously described (Tas- stages of regeneration. Regenerating newt limbs were exam- sava and Acton, 1989). ined at 3, 5, 7, 10, and 14 days after amputation (pre-blastema stages) and at early-bud, mid-bud, late-bud, palette, and digit Antibodies stages of regeneration (Iten and Bryant, 1973). Regenerates mAb MT1 (matrix 1; formerly called 4G3; Goldhamer and were always sampled so that a 1-2 mm portion of stump tissue Tassava, 1986; Tassava, 1988) was obtained by immunizing was included. mice to homogenates of mid- and late-bud blastemas of adult To test whether mAb MT1 might be reacting to tenascin or newts. The immunization protocol and survey method for fibronectin, three different tests were performed. First, we putative mAbs were similar to those described for mAb WE3 compared the reactivity patterns of mAb MT1, pAbTN, and (Tassava et al. 1986). mAb MT1 was precipitated with pAbFN in adjacent cryostat sections of newt blastemas and ammonium sulfate from hybridoma culture medium, dis- limb buds. Second, a double labeling experiment was carried solved in phosphate-buffered saline (PBS), dialyzed against out on the same sections of newt and axolotl blastemas. PBS, and stored frozen as a stock solution. The stock solution Sections were incubated sequentially in mAb MT1, FITC- could be diluted over 1000 x before loss of reactivity in tissue labeled goat anti-mouse IgG (Cappel), pAbTN and rhoda- sections. We determined that mAb MT1 is an IgM antibody mine-labeled goat anti-rabbit IgG (affinity purified; Bio- (Isotyping kit; Calbiochem). Preliminary data on mAb MT1 Rad). Each step consisted of a lh incubation followed by reactivity have been reported (Goldhamer and Tassava, 1986; washes as described. The specificity of secondary antibodies Tassava, 1988). and of optical filters was confirmed by including controls in The rabbit polyclonal antiserum against chicken tenascin which only a single pair of Abs was applied. Double-labeled (pAbTN) was a kind gift from Dr R. Chiquet-Ehrismann, and sections were viewed by indirect immunofluorescence with Matrix antigen of regenerating limb blastemas 659 the appropriate filter combinations so that F1TC and rhoda- PBS-EDTA, and resuspended in SDS sample buffer without mine could be examined independently. Third, a competition /3-mercaptoethanol, and boiled for lOmin. The beads were experiment on tissue sections was designed in which the pelleted by centrifugation and proteins in the supernatant ability of pAbTN and pAbFN to inhibit binding of mAb MT1 were separated by SDS-PAGE as above. For reduced con- was tested. Sections of newt blastemas were first incubated in ditions, /S-mercaptoethanol was added to the supernatant to a 5% non-fat dry milk in PBS for30min. Either pAbTN (1:10 final concentration of 10% before SDS-PAGE. Immuno- dilution in PBS) or pAbFN (no dilution) was next applied to reactivities to mAb MT1, pAbTN and pAbFN of the immuno- separate sections and allowed to react for 1 h after which mAb precipitates were assessed by immunoblotting as above. MT1 was applied to the same sections without intervening washes. After a lh incubation, and subsequent washes, secondary Ab specific to mAb MT1, rhodamine-conjugated Results goat anti-mouse IgM, was then applied as above, and sections were examined for immunoreactivity. Controls in which the Distribution of MT1 antigen in the limb stump secondary Ab was applied without prior incubation in mAb In the limb stumps of axolotls and adult newts, mAb MT1 showed no detectable reactivity against either pAbTN or MT1 reacted to tendons, myotendinous junctions, pAbFN. perichondrium/periosteum, a thin layer at the junction Immunoblotting between integumentary glands and epidermis (in Both MT1 antigen and tenascin could be extracted from newts), and to granules within the basal layers of the various tissues using a high pH buffer (Riou et al. 1988; epidermis (Fig. 1A-C, 2C, D; Table 1). Since reactivity Crossin et al. 1986). Briefly, mid- to late-bud blastemas of patterns to mAb MT1 were similar in axolotls of all axolotls were homogenized in a frosted glass homogenizer three sizes, only data from 35 mm larvae will be with 10 volumes of 30 mM diethylamine (pH11.5), 1 ITIM reported here. Reactivity in the limb stump of the EDTA, 2mM phenylmethylsulphonyl fluoride (PMSF), and axolotl was similar to that of newts except that the 2\M leupeptin and extracted on ice for 6h with shaking. The perichondrium of axolotls stained more strongly than extracts were cleared by centrifugation at 10000 g for lOmin the periosteum of newts; also, the granular reactivity in at 4°C, dialyzed against dd H2O and aliquoted for immediate use or for storage at -80°C. axolotl epidermis was not as abundant as in newts Extracts were mixed with an equal volume of 2x SDS (Table 1; see newt epidermal reactivity in Fig. 2C, D). sample buffer (Laemmli, 1970) with or without /S-mercapto- While the present results do not exclude intracellular ethanol. Samples were separated on 5% (reducing con- reactivity, the majority of mAb MT1 reactivity in the ditions) or 3 % (nonreducing conditions) discontinuous unamputated limb appeared to be extracellular. In SDS-polyacrylamide gels with 2.5 % stacking gels (Laemmli, newts the reactivity linking glands to the epidermis 1970). After gel electrophoresis, proteins were transferred to (Fig. 2C, D) and also the majority of reactivity of nitrocellulose (Towbin et al. 1979) and then blocked for tendons (Fig. 1C) was most likely extracellular. In 30min with 5% non-fat dry milk in Tris-buffered saline sections of both axolotl and newt epidermis which were (TBS). Strips of nitrocellulose were incubated separately in pAbTN (1:500), pAbFN (1:500), nonimmune rabbit serum examined both by epifluorescence (Fig. IB) and (1:100), nonimmune mouse serum (1:100), or mAb MT1 Nomarski optics (not shown), the mAb MT1 reactive (1:100), diluted with 5 % non-fat dry milk in TBS, overnight granules appeared in most cases to outline cells and at room temperature with gentle agitation. Strips were therefore might largely be extracellular. washed in TBS with 0.05% Tween-20 three times before incubation in appropriate horseradish peroxidase-conjugated Expression of MT1 antigen during limb regeneration secondary antibodies (goat anti-rabbit IgG for pAbTN, In the regenerating limb, changes in mAb MT1 reac- pAbFN, and nonimmune rabbit serum; goat anti-mouse IgG tivity appeared early, already at 1 day after amputation for nonimmune mouse serum; goat anti-mouse IgM (for mAb MT1). After washing as above, bands were visualized by in 35 mm axolotls and 5 days after amputation in the developing the strips in a solution of 4-chloro-l-naphthol adult newt (Table 2). Except for timing of appearance containing H2O2, and photographed after color development. 3 1 Molecular weight standards were myosin (MT 440 xlO Table 1. mAB MT1 reactivity in unamputated limb 3 nonreduced, M, 205 xlO reduced), /3-galactosidase (Mr 116 3 3 tissues of newts and axolotls xlO ), phosphorylase b (Afr 97 xlO ), and bovine plasma albumin (M 66 xlO3). Relative amount r Tissues/Cells examined of reactivity Immunoprecipitations Epidermis Epidermis/gland junction Immunoadsorbents were prepared by coupling mAb MT1 to Tendons the activated ester agarose matrix, Affi-Gel 15 (BioRad), at Myotendinous junctions approximately 10 mg antibody per lml of beads. To an Perichondrium (axolotls) aliquot of extract used for immunoblotting, sodium phosphate Periosteum (newts) (dibasic), NaCI and EDTA were added to a final concen- Integumentary glands tration of 10mM phosphate, 150mM NaCI, and lmM EDTA, Muscle pH7.4 (PBS-EDTA). Extract was first incubated with Dennis quenched Affi-Gel 15 for 1 h and centrifuged to remove any Nerve proteins nonspecifically bound to the beads. The preadsorbed Blood vessels extract was incubated with the mAb MT1 coupled Affi-Gel 15 overnight at 4°C with gentle shaking. Immunoprecipitates 'Assessed by examining immunofluorescence on sections on a were collected by centrifugation, washed three times with scale from 0 (-) to 3 (+ + + ). 660 H. Onda, D. J. Goldhamer and R. A. Tassava

Fig. 2. Indirect immunofluorescence micrographs illustrating reactivity of mAb MT1 in the regenerate and the stump of the axolotl and newt. (A) Longitudinal section of the limb stump and early bud blastema of a larval axolotl. Reactivity of mAb MT1 can be seen in the stump perichondrium (arrows), in the area of dedifferentiation along the radius (r), in the blastema (b) and in a layer underneath the wound epithelium (arrow heads). Reactivity in the epidermis and wound epithelium cannot be seen at this low magnification. The dotted line indicates the level of amputation. Bar, 100/.an. (B) Longitudinal section of the distal tip of a mid-bud blastema of the adult newt. The reactivity of mAb MT1 is intense and uniformly distributed among blastema mesenchyme cells. Note that the intensely reactive layer of MT1 underneath the wound epithelium (arrow heads) seems to be continuous with the mAb MT1 reactivity in the mesenchyme by thin strands of mAb MT1 reactive material. Some mAb MT1 reactive granules can be seen in the wound epithelium and some of the cells of the wound epithelium seem to contain a diffuse intracellular material that is MT1 positive (arrows). Bar, 200//m. (C) Longitudinal section of a 14 day post-amputation, pre- blastema stage, newt regenerate. Considerable reactivity to mAb MT1 is seen among the cells in the area of dedifferentiation which drops off suddenly just proximal to the level of amputation (dotted line). The layer of reactivity underneath the wound epithelium is not as defined as in later stages (compare Fig. 2B) and the distal tip of the wound epithelium (arrows) is more reactive to mAb MT1 than the rest of the wound epithelium. More obvious granular reactivity can be seen in the epidermis proximal to the level of amputation (arrow heads) and mAb MT1 reactivity is strong at the junction of glands (g) and epidermis. This section grazes the distal radius (r) so that only a small amount of periosteum (p) is visible. Bar, 200/on. (D) A higher magnification of a longitudinal section through the border between the stump and early-bud blastema of an adult newt. The strong granular reactivity in Fig. 1. Indirect immunofluorescence micrographs the epidermis (arrow heads) decreases markedly near the illustrating reactivity of mAb MT1 in the unamputated limb level of amputation (dotted line) and in the wound and stump of the larval axolotl. (A) Cross-section through epithelium (we), near the point where MT1 reactivity the level of mid radius-ulna of the unamputated axolotl begins to appear in a layer underneath the wound forelimb. Strong reactivity of mAb MT1 can be seen in the epithelium (arrows). Some of the thick and thin mAb MT1 perichondrium (p), tendons (t) and myotendinous junctions reactive cords seem to be organized in a proximal-distal (j). Other tissues, including muscle (m) and cartilage (c), do direction at the stump-regenerate border (see also Fig. 2C). not react to mAb MT1 in unamputated limbs. Bar, 100 /.an. Bar, 100/.an. (E) Longitudinal section of the distal end of a (B) Longitudinal section through the skin of the limb stump digit stage regenerate of a larval axolotl. mAb MT1 of an axolotl proximal to the level of amputation. At this reactivity is still strong distal to the amputation level (left magnification, fine granular reactivity outlining epidermal edge of the photograph) in areas of soft tissue development cells can be seen. Reactivity is usually strongest in the basal and among the blastema mesenchyme (b) at the still layers of the skin epidermis (right side of the photograph), growing tips of the digits (arrow heads). Newly and is weakly present or absent in the outer layers. Bar, differentiating cartilage of the digits (d) shows mAb MT1 100 /nn. (C) A higher magnification of a cross section of the reactivity in the matrix; although the reactivity cannot be unamputated axolotl forearm. Note the strongly reactive distinguished clearly at this magnification, differentiating tendon (t) connecting the also reactive perichondrium (p) chondrocytes show diffuse intracellular reactivity. Note that and myotendinous junction (j). Muscle (m) is non-reactive. differentiating cartilage (c) and muscles (m) lose mAb MT1 See Fig. 1A. Bar, 15 ^m. reactivity more proximally. Bar, 100 f.an. and disappearance, reactivity in axolotls and newts was reactivity was very strong and relatively uniformly similar. Staining first appeared as a fine band under the distributed throughout the blastema matrix and in a wound epithelium and as fibers and cords among thick band under the wound epithelium (Fig. 2B-D). muscle and connective tissues in the area of dedifferen- During pre-blastema and blastema stages of regener- tiation (Fig. 2A, C). Reactivity became stronger as the ation, the band of MTl-reactive material under the number of blastema cells increased, so that through wound epithelium ended abruptly at the level of ampu- early-, mid-, and late-bud blastema stages, mAb MT1 tation (Fig. 2A, C, D). The thick band of reactive Matrix antigen of regenerating limb blastemas 661 662 H. Onda, D. J. Goldhamer and R. A. Tassava

Table 2. Time of appearanceand distribution of mAb MTl reactivity during adult newt limb regeneration Days after amputation/stages:> Tissues/Cells examined l 5 10 14 EB MB LB PAL DIG Wound epithelium - +/— +/- +/ — Wound epi/ - + ++ + + mesenchyme border Area of - + ++ + + dedifferentiation Blastema mesenchyme N3 N N N matrix Cartilage matrix4 N N N N N N Chondrocytes4 N N N N N N N + + (intracellular) Nerves ______Bloodvessels NNNN-

1 Assessed by examining immunofluorescence on sections on a scale from 0 (-) to 3 ( + ++). 2EB=early-bud; MB=mid-bud; LB=late-bud; PAL=palette; DIG=digit (stages according to Iten and Bryant, 1973). 3(N) indicate tissues/cells examined are not present in the regenerate at this stage. 4The reactivities in the early differentiating cartilage were examined. Neither matrix nor chondrocytes of fully differentiated cartilage reacted to mAB MTl (see text).

material subjacent to the wound epithelium, as well as blastemas and limb stumps and that the MTl antigen is the majority of staining throughout the blastema, was distributed in an almost identical fashion. Unlike extracellular, but a low level of intracellular reactivity pAbTN, fibronectin polyclonal Ab (pAbFN) reacted cannot be ruled out (Fig. 2B, D). At differentiation more diffusely in the blastema and was weakly reactive stages, MTl reactivity remained high in mesenchyme of to the layer underneath the wound epithelium (not the distal tip of the regenerate, under the distal wound shown). In the stump, pAbFN reactivity was more epithelium at the tips of digits, and in the perichon- distinct from the reactivity of mAb MTl and pAbTN. It drium, and was somewhat less intense but present reacted to basement membranes of muscle and epider- within the area of differentiating chondrocytes; reac- mis, dermis, and perineurium of nerves (not shown), as tivity decreased in differentiating muscle and connec- previously described (Repesh et al. 1982; Gulati et al. tive tissues (Fig. 2E), except in tendons, where reac- 1983). tivity persisted (not shown). As differentiation Double labeling on the same section revealed near proceeded, the reactive band under the wound epi- identical distributions of mAb MTl (FITC-labeled thelium was lost proximally near the level of ampu- secondary Ab) and pAbTN (rhodamine-labeled sec- tation (Fig. 2E). ondary Ab) reactivity, even to some of the finest strands Throughout regeneration, the wound epithelium was of reactive matrix (Fig. 4A, B). Finally, a competition less reactive to mAb MTl than skin epidermis in both experiment further suggested that the MTl antigen newts and axolotls but some granular reactivity was might be tenascin. Prior incubation of sections with nevertheless present (Fig. 2B, C). In some sections, a pAbTN almost totally eliminated subsequent binding of diffuse reactivity was sometimes seen within some cells mAb MTl; however, prior incubation with undiluted of the wound epithelium nearest the distal mesenchyme pAbFN did not decrease subsequent binding of mAb (Fig. 2C), especially at early blastema stages, but this MTl (Fig. 5A, B, C). was not consistent and requires further study. Distribution of MTl antigen and tenascin in Colocalization of tenascin with MTl antigen developing limbs A polyclonal Ab to chicken tenascin (pAbTN) reacted Since tenascin exhibits a spatially restricted distribution to sections of both axolotls and newts in a fashion in axolotl embryos (Epperlein etal. 1988), we examined similar to that of mAb MTl. It was of interest therefore both axolotl and newt embryos for mAb MTl reac- to determine whether mAb MTl might in fact be tivity. Embryos of newts and axolotls exhibited similar reacting to urodele tenascin. Very similar patterns were patterns of reactivity to mAb MTl (Fig. 6) and both seen when the reactivities of mAb MTl and pAbTN resembled closely the reactivity reported for pAbTN were compared on adjacent sections (Fig. 3A, B). (Epperlein et al. 1988), although the latter study did not Reactivities of both mAb MTl and pAbTN were seen include limb bud. Reactivity of mAb MTl was uni- among the blastema mesenchyme and in a layer under- formly present amongst the mesenchyme throughout neath the wound epithelium. Both Abs exhibited a the forelimb and hindlimb buds, and in a layer under granular reactivity in skin epidermis (to a lesser degree the distal ectoderm (Fig. 6A). The developing limb bud in wound epithelium) and a fibrous reactivity in matrix, including the layer under the ectoderm, was periosteum/perichondrium (not shown). These results reactive at all stages examined, beginning at the time show that tenascin is present in both newt and axolotl the limb bud was first recognizable as an outgrowth. Matrix antigen of regenerating limb blastemas 663

Fig. 3. Indirect immunofluorescence micrographs illustrating the similar patterns of (A) mAb MTl and (B) pAbTN reactivities in adjacent, longitudinal cryostat cut sections of an adult newt late-bud stage blastema. The dark area in (A) is a tear in the section (asterisk). Bar, 200 ^m.

mAb MTl reactivity was also seen outlining the notochord and the neural tube, in association with the somites, and in the connective tissue of the dorsal fin (Fig. 6B). Confirmation of other reactivities in the embryo will require more study. Finally, a double Ab test of mAb MTl and pAbTN on limb bud sections (as above for blastemas) revealed essentially identical patterns of reactivity for the two Abs (not shown).

Immunoblot analysis of blastema extracts In order to determine the molecular nature of the MTl antigen and to compare it with the molecule identified by pAbTN in newts and axolotls, immunoblot analyses of both newt and axolotl blastema extracts were con- ducted. Only data from axolotls are shown (Fig. 7) since identical results were obtained when newt blastema extract was analyzed. We observed previously that MTl reactivity is lost when newt or axolotl blastema extracts were treated Fig. 4. Indirect immunofluorescence micrographs with reducing reagents such as /J-mercaptoethanol illustrating double labeling patterns of mAb MTl and (Goldhamer and Tassava, unpublished data). This re- pAbTN on a section of an axolotl limb blastema. Both Abs sult was confirmed here by immunoblotting of axolotl reacted to a fine fibrous matrix material around blastema blastema extract under reducing conditions (Fig. 7A, cells (arrow heads) and to a thick cord-like material lane 2). The same sample stained with pAbTN showed underneath the wound epithelium (arrows). (A) A section 3 a series of strongly reactive bands of Mr 210-250 xlO , of mid-bud stage blastema of an axolotl photographed 3 3 under fluorescein optics to visualize mAb MTl reactivity. and less intense 105X10 and 80X10 bands (Fig. 7A, (B) The same field photographed under rhodamine optics lane 1). These results indicate that disulfide bonds are to visualize pAbTN reactivity. The area photographed is the essential for the maintenance of antigenicity to mAb distal tip of the blastema, at the border of wound MTl but not to pAbTN. The range of subunit molecu- epithelium (we) and mesenchyme (m). Bar, 15 jrni. lar weights from 210-250K are in good agreement with 664 H. Onda, D. J. Goldhamer and R. A. Tassava

col (Fig. 7A, lane 1). pAbFN reacted to three distinct 3 3 3 bands of Mr 250X10 , 235xlO , and 220X10 (Fig. 7A, lane 3). 220K is the reported relative molecular weight of the fibronectin subunit, which has almost an identical migration rate on SDS gels as the major subunit of chicken tenascin (Chiquet and Fambrough, 19846). The origin of the two larger bands reactive to pAbFN is currently unknown; however, based on reported vari- ations of fibronectin subunit mobilities on SDS gels (Hynes, 1985), it is possible that the two larger bands reflect heterogeneity of blastema fibronectin. Neither nonimmune rabbit nor mouse serum showed reactive bands corresponding to tenascin or fibronectin subunits (not shown). Nonreduced MT1 antigen was sufficiently resolved in a 3 % polyacrylamide gel so as to detect two distinct bands which migrated into the gel only a short distance (Fig. 7B, lane 2). Precise molecular weight determi- nation of these nonreduced molecules is difficult. How- ever, from comparison to available molecular weight standards, the larger band corresponds to roughly MT lOOOxlO3, which is the estimated molecular weight of the intact oligomer of tenascin (Vaughan et al. 1987). In fact, bands of similar molecular weight reacted to pAbTN (Fig. 7B, lane 1). Heterogeneity of subunit composition has been described for tenascin (Chiquet and Fambrough, 1984a,b). The two high molecular weight polypeptides revealed by mAb MT1 may rep- resent tenascin heterogeneity among the blastema cell population. Alternatively, the smaller band could simply be a proteolytic fragment of the larger one. In addition to the high molecular weight oligomer, pAbTN identified several smaller polypeptides which did not react to mAb MT1. The sizes of these bands correspond to dimeric and monomeric forms of tenascin (Riou et al. 1988). The result can be interpreted as either proteolysis occurring during the extraction as suggested by Riou et al. (1988) or existence of incom- pletely assembled molecules of tenascin in the tissues. 3 3 3 pAbFN reacted to Mr 500xlO , 440xlO , and 235xlO bands of the nonreduced sample (Fig. 7B, lane 3). Fig. 5. Indirect immunofluorescence micrographs Again, presence of a 500K band, which is larger than illustrating reactivity to mAb MT1 after first incubating the expected dimer of 440K, may be related to subunit sections of mid-bud stage newt blastemas with 5 % non-fat dry milk alone (A), with pAbFN (B), or with pAbTN (C). size heterogeneity in blastema fibronectin. Neither of Sections were subsequently incubated with mAb MT1, and the control sera reacted to MT1, TN, or FN specific then rhodamine-labeled secondary antibody against mAb bands (not shown). MT1. These micrographs are at the level of amputation. Note that mAb MT1 reactivity is much reduced, i.e. Immunoprecipitation of tenascin using mAb MT1 competed, by prior incubation of the section with pAbTN As shown in Fig. 8A, mAb MT1 precipitated proteins (C), but mAb MT1 staining is not diminished by prior that react to pAbTN (lane 1) but not to pAbFN (lane 3) incubation with non-fat dry milk alone (A), or with pAbFN after reduction. Under nonreducing conditions (B). Sections were photographed using constant exposure (Fig. 8B), the immunoprecipitates were recognized by time. Bar, 50 j/m. both pAbTN and mAb MT1 but not by pAbFN. The patterns of the reactivities were essentially identical to the subunits of tenascin from chicken (Chiquet and the results of the immunoblots with crude extract Fambrough, 19846), Xenopus laevis (Epperlein et al. except that there was no detectable fibronectin present 1988), and Pleurodeles waltlii (Riou et al. 1988). Since in the precipitate. The presence of the smaller pAbTN crude extract of blastema prepared directly in SDS reactive bands which did not react to mAb MT1 sample buffer did not show 105K and 80K bands (not indicates that these peptides were recognized by mAb shown), they are likely proteolytic fragments of the MT1 during immunoprecipitation but not after above subunits due to the prolonged extraction proto- SDS-PAGE. A possible explanation of this apparently Matrix antigen of regenerating limb blastemas 665

Fig. 6. Indirect immunofluorescence micrographs illustrating mAb MTl reactivity in axolotl embryos. (A) Longitudinal cryostat cut section of an early forelimb bud. mAb MTl reactivity is in the bud mesenchyme (m) and in a layer underneath the bud ectoderm (arrows). Axolotl hind limb buds as well as newt limb buds exhibit similar reactivity to mAb MTl. Bar, 50 jim. (B) Cross-section of an axolotl embryo at the level of the tail fin. mAb MTl reactivity can be seen in the tail fin matrix (fm), extending dorsally from the neural tube, in a layer around the neural tube (nt), around the notochord (nc), and in the early developing musculature (mu) lateral to the neural tube. A layer of mAb MTl reactivity is present under the ectoderm of the dorsal fin but not along the body wall under ectoderm lateral to the developing musculature. Bar, 100/

S - Fig. 7. Immunoblot analyses of axolotl s - limb blastema extracts. High pH extracts from axolotl limb blastemas were mixed with 2x SDS sample buffer with or without /3-mercaptoethanol. After gel electrophoresis, proteins were transferred 205- to nitrocellulose, and stained with pAbTN (lane 1), mAb MTl (lane 2), or pAbFN 440- (lane 3). (A) Reduced sample run on a 5 % polyacrylamide gel. Note strongly reactive 210-250K bands revealed by pAbTN 116- (bracket) but absence of MTl antigen staining. (B) Non-reduced sample run on a 97- 3% polyacrylamide gel. Both pAbTN and 66- 205- mAb MTl reacted to a 1000K band (arrows) corresponding to the molecular weight of intact tenascin (see text). Positions of molecular weight standards 3 (MT xl(T ) and start of running gels (S) 1 2 are shown at left.

epithelium during the 2nd week post-amputation in Restricted expression of MTl antigen in the limb newts (Tassava et al. 1986). It may be significant that stump those wound epithelial cells that react most strongly to The MTl antigen has a limited distribution in the limb mAb WE3 reside at the base of the epithelium (Tassava stump, where its presence is indicated by mAb MTl et al. 1986), adjacent to the strongly reactive MTl layer reactivity to perichondrium and periosteum, tendons, between the wound epithelium and mesenchyme. The myotendinous junctions, gland-epidermis junctions and correspondence between the timing and distribution of epidermis. It could be speculated that the MTl antigen reactivity to mAb MTl and those of other mAbs acts as an adhesive agent, holding tendons to muscle reactive during regeneration is an area worthy of and glands to epidermis. In the epidermis, mAb MTl further study. reacts to a granular component, present largely between the cells within the non-differentiated, basal A B layers. The role for the MTl antigen in epidermis is at present unknown. Immunological relationship of MTl antigen with tenascin The possibility was considered that mAb MTl was OP reacting to one of the extracellular matrix components previously described in regenerating limbs (Tassava, 1988). However, the reactivity patterns of mAb MTl in the unamputated limb and in the blastema were not what would be expected of the commonly known extracellular matrix components such as collagens, 1 2 3 1 2 3 laminin, fibronectin and hyaluronic acid, all of which have wider distributions (Repesh et al. 1982; Gulati et Fig. 8. Immunoprecipitation of tenascin from axolotl al. 1983; Mescher and Munaim, 1986). On the other blastema extract using mAb MTl. Immunoprecipitated hand, tenascin, a relatively recently characterized proteins were resuspended in SDS sample buffer with or extracellular matrix glycoprotein (Vaughan et al. 1987), without /3-mercaptoethanol. (A) Reduced became a strong candidate based on comparison of its immunoprecipitates run on a 5 % polyacrylamide gel. temporal and spatial distribution with that of mAb MTl (B) Non-reduced immunoprecipitates run on a 3 % reactivity. Consider for example that tenascin has been polyacrylamide gel. The proteins were transferred to shown to be present in tendons, myotendinous junc- nitrocellulose, and stained with pAbTN (lane 1), mAb MTl (lane 2), or pAbFN (lane 3). The patterns of the reactivities tions and perichondrium (Chiquet and Fambrough, were essentially identical to the results of the immunoblots 1984a), each of which is also reactive to mAb MTl. (Fig. 7) except for the absence of FN staining. Based strictly on Ab reactivity patterns, it was reason- Matrix antigen of regenerating limb blastemas 667 able to further investigate whether the MT1 antigen is (Chiquet and Fambrough, 1984a, b), epithelial— tenascin or is an antigen closely associated with ten- mesenchymal interactions during kidney and mammary ascin. Since tenascin is known to partially colocalize development (Inaguma et al. 1988), cartilage and bone with fibronectin in vivo (Chiquet and Fambrough, differentiation (Mackie et al. 1987), and mammary 1984a), and copurify in cell-surface fibronectin prep- tumor development (Inaguma et al. 1988). In cell arations (Chiquet-Ehrismann etal. 1986), all the exper- culture, tenascin is mitogenic for mammary tumor cells iments performed to identify the MT1 antigen included (Chiquet-Ehrismann etal. 1986) and inhibits cell attach- tests for fibronectin. All the results indicate that the ment to fibronectin (Chiquet-Ehrismann et al. 1988). MT1 antigen is distinct from fibronectin. Finally, fetal bovine serum and transforming growth Evidence that the MT1 antigen is tenascin was factor (5 (TGF ft) individually stimulate synthesis of provided from immunocytochemical comparisons of tenascin by human and chick embryo fibroblasts (Pear- mAb MT1 reactivity with that of pAbTN. In embryos, son et al. 1988). In light of these findings, and consider- mAb MT1 reacted to somites, to a layer surrounding ing the suggestion of Pearson et al. (1988) that tenascin the neural tube, and to the matrix of the developing tail may be important to wound healing and regeneration fin. This pattern of reactivity resembled that reported (see Mustoe et al. 1987), it is worth investigating for tenascin in chicken and amphibian embryos (Epper- possible roles for tenascin in limb regeneration. The lein et al. 1988). In adjacent sections of blastemas, the distribution of tenascin, near or in contact with essen- two Abs exhibited very similar reactivity patterns tially every mesenchymal cell within the blastema, within the blastema and the limb stump. Double could mean it is mitogenic or assists somehow in the labeling with mAb MT1 and pAbTN on the same mitogenic action of other molecules. The layer of blastema section also revealed essentially identical tenascin under the wound epithelium could be relevant reactivity patterns. Furthermore, pAbTN competed to the interaction between wound epithelium and mes- specifically with mAb MT1 binding to tissue sections enchyme. whereas pAbFN did not. Previous studies imply strongly that tenascin is syn- Biochemical characterization provided further evi- thesized strictly by mesenchyme (Inaguma et al. 1988). dence that the MT1 antigen corresponds to urodele We show here that both mAb MT1 and pAbTN exhibit tenascin. Tenascin is a large, six-armed molecule, held a granular reactivity in both newt and axolotl epidermis together by disulfide bonds (Vaughan et al. 1987). On and to a lesser extent in the wound epithelium; how- immunoblots, mAb MT1 and pAbTN reacted to high ever, we do not know the cellular origin of this granular molecular weight bands of identical size under nonre- reactivity. In preliminary studies (Tassava, unpublished ducing conditions, but with reduced samples, mAb data), mAb MT1 has been shown to react more strongly MT1 reactivity was lost. Thus, the MT1 antigen also to basal layers of specialized epidermis, such as the contains disulfide bonds, and a comformation depen- nuptial pads of the male newt hindlimb, compared to dent antigen is lost upon reduction. Finally, mAb MT1 the same region of female newts, or to other areas of precipitated a high molecular weight protein corre- epidermis of the male. These studies are being ex- sponding in size to oligomeric tenascin, which was tended and may provide clues as to what role tenascin recognized by both pAbTN and mAb MT1, but not might have in the epidermis, whether the granules are pAbFN. Under reducing conditions, this immuno- extra- or intracellular, and their cellular origin. precipitate lost reactivity to mAb MT1 but maintained Given the many developmental phenomena occur- reactivity to pAbTN. ring during regeneration, including dedifferentiation, Based on the immunocytochemical data, immuno- cellular proliferation, epithelial-mesenchymal interac- blot analyses, and immunoprecipitation results, it tions, nerve ingrowth and pattern formation, it is likely seems reasonable to suggest that mAb MT1 reacts to that many different matrix components play interactive the intact molecule of newt and axolotl tenascin and roles. Not surprisingly then, previous studies have that tenascin is an important component of the blas- shown that fibronectin (Repesh et al. 1982; Gulati et al. tema matrix. Formal proof that MT1 antigen is urodele 1983), laminin (Gulati et al. 1983), and hyaluronic acid tenascin requires complete biochemical characteriz- (Smith etal. 1975; Mesher and Munaim, 1986; Mescher ation and elucidation of its gene sequence. In this and Cox, 1988), might be important to regeneration. In regard, purifying the MT1 antigen from axolotl tissue this regard, Mescher and Cox (1988) found a corre- extracts by immunoaffinity chromatography and lation between hyaluronate accumulation, re-inner- screening a newt blastema cDNA library with a chicken vation, and blastema formation in axolotls. On the tenascin cDNA probe will add further data relevant to other hand, Maden and Keeble (1987), in examining the identity of the MT1 antigen. retinoic acid (RA) induced distal-proximal dupli- cation, found no alteration in fibronectin distribution or Functional roles of tenascin during regeneration synthesis compared to control regenerates. We have Strong immunoreactivity to pAbTN shows that tenascin begun studies with mAb MT1 and other matrix-reactive is abundant in the blastema and limb bud of newts and mAbs to determine to what extent expression of these axolotls. While tenascin has not previously been investi- matrix antigens are influenced by RA, nerves and/or gated during regeneration or early development of hormones. limbs, a variety of evidence has implicated this large matrix glycoprotein in muscle/tendon morphogenesis The mAb MT1 was obtained while R.A.T. was a senior 668 H. Onda, D. J. Goldhamer and R. A. Tassava

research fellow in the Developmental Biology Laboratory, ribonucleic acid synthesis during the initiation of limb Massachusetts General Hospital, Dr Jerome Gross, Director. regeneration in larval axolotls (Ambystoma mexicanum). J. exp. Efforts at the MGH were supported by NIH grant AM3564 Zool. 185, 45-54. and EY2252 to J. Gross and senior research fellowship award KJNTNER, C. R. AND BROCKES, J. P. (1984). Monoclonal antibodies identify blastema cells derived from dedifferentiating muscle in F33HD06675-1 to R. Tassava. Studies at The Ohio State newt limb regeneration. Nature, Loud. 308, 67-69. University were supported by NIH grant HD22024 to R.A.T. LAEMMLI, U. K. (1970). Cleavage of structural proteins during the D.J.G. and H.O. were supported by fellowships from The assembly of the head of bacteriophage T4. Nature, Loud. 227, Ohio State University during a portion of this work. 680-689. MACKIE, E. J., THESLEFF, I. AND CHIQUET-EHRISMANN, R. (1987). References Tenascin is associated with chondrogenic and osteogenic differentiation in vivo and promotes chondrogenesis in vitro. J. Cell Biol. 105, 2569-2579. AUFDERHEIDE, E., CHIQUET-EHRISMANN, R. AND EKBLOM, P. MADEN, M. AND KEEBLE, S. (1987). The role of cartilage and (1987). Epithelial-mesenchymal interactions in the developing fibronectin during respecification of pattern induced in the kidney lead to expression of tenascin in the mesenchyme. J. Cell regenerating amphibian limb by retinoic acid. Differentiation 36, Biol. 105, 599-608. 175-184. BOUCAUT, J. C. AND DARRIBERE, T. (1983). Fibronectin in early MAIER, C. E., WATANABE, M., SINGER, M., MCQUARRIE, I. G., amphibian embryos: Migrating mesodermal cells contact SUNSHINE, J. AND RUTISHAUSER, U. (1986). Expression and fibronectin established prior to gastrulation. Cell Tissue Res. function of neural cell adhesion molecule during limb 234, 135-145. regeneration. Proc. natn. Acad. Sci. U.S.A. 86, 8395-8399. CHALKLEY, D. T. (1954). A quantitative histological analysis of MESCHER, A. L. AND COX, C. A. (1988). Hyaluronate accumulation forelimb regeneration in Trirurus viridesceiis. J. Morph. 94, and nerve-dependent growth during regeneration of larval 21-70. Ambystoma limbs. Differentiation 38, 161-168. CHIQUET, M. AND FAMBROUGH, D. M. (1984a). Chick myotendinous MESCHER, A. L. AND MUNAIM, S. I. (1986). Changes in the antigen. I. A monoclonal antibody as a marker for tendon and extracellular matrix and glycosaminoglycan synthesis during the muscle morphogenesis. /. Cell Biol. 98, 1926-1936. initiation of regeneration in adult newt forelimbs. Anat. Rec. CHIQUET, M. AND FAMBROUGH, D. M. (19846). Chick myotendinous 214, 424-431. antigen. II. A novel extracellular glycoprotein complex consisting MESCHER, A. L. AND TASSAVA, R. A. (1975). Denervation effects of large disulfide-linked subunits. J. Cell Biol. 98, 1937-1946. on DNA replication and mitosis during the initiation of limb CHIQUET-EHRISMANN, R., KALLA, P., PEARSON, C. A., BECK, K. regeneration in adult newts. Devi Biol. 44, 187-197. AND CHIQUET, M. (1988). Tenascin interferes with fibronectin MUSTOE, T. A., PIERCE, G. F., THOMASON, A., GRAMATES, P., action. Cell S3, 383-390. SPORN, M. B. AND DEUEL, T. F. (1987). Accelerated healing of CHIQUET-EHRISMANN, R., MACKIE, E. J., PEARSON, C. A. AND incisional wounds in rats induced by transforming growth factor SAKAKURA, T. (1986). Tenascin: an extracellular matrix protein p. Science, Wash. DC. 237, 1333-1336. involved in tissue interactions during fetal development and PEARSON, C. A., PEARSON, D., SHIBAHARA, S., HOFSTEENGE, J. AND oncogenesis. Cell 47, 131-139. CHIQUET-EHRISMANN, R. (1988). Tenascin: cDNA cloning and CROSSIN, K. L., HOFFMAN, S., GRUMET, M., THIERY, J. P. AND induction by TGF-/3. EMBO J. 7, 2977-2981. EDELMAN, G. M. (1986). Site-restricted expression of cytotactin REPESH, L. A., FITZGERALD, T. J. AND FURCHT, L. T. (1982). during development of the chicken embryo. J. Cell Biol. 102, Changes in the distribution of fibronectin during limb 1917-1930. regeneration in newts using immunocytochemistry. EPPERLEIN, H. H., HALFTER, W. AND TUCKER, R. P. (1988). The Differentiation 22, 229-240. distribution of fibronectin and tenascin along migratory pathways Riou, J. F., SHI, D. L., CHIQUET, M. AND BOUCAUT, J. C. (1988). of the neural crest in the trunk of amphibian embryos. Expression of tenascin in response to neural induction in Development 103, 743-756. amphibian embryos. Development 104, 511-524. GOLDHAMER, D. J. AND TASSAVA, R. A. (1986). A blastema specific SMITH, G. N., TOOLE, B. P. AND GROSS, J. (1975). Hyaluronidase antigen; reactivity to mAb 4G3. Amer. Zool. 26, 57A. activity and glycosaminoglycan synthesis in the amputated newt GOLDHAMER, D. J., TOMUNSON, B. L. AND TASSAVA, R. A. (1989). limb: Comparison of denervated, non-regenerating limbs with A developmentally regulated wound epithelial antigen of the regenerates. Devi Biol. 43, 221-232. newt limb regenerate is also present in a variety of TASSAVA, R. A. (1988). Limb regeneration in newts: an secretory/transport cell types. Devi Biol. 135, 392-404. immunological search for developmentally significant antigens. In GORDON, H. AND BROCKES, J. P. (1987). Appearance and Proc. 6th M. Singer symposium (ed. S. Inoue etal.), pp. 69-78. regulation of an antigen associated with limb regeneration in Maebashi: Okada Printing. Notophthalmus viridcscens. J. exp. Zool. 247, 232-243. TASSAVA, R. A. AND ACTON, R. D. (1989). Distribution of a wound GRILLO, H. C, LAPIERE, C. M., DRESDEN, M. H. AND GROSS, J. epithelium antigen in embryonic tissues of newts and (1968). Collagenolytic activity in regenerating forelimbs of the salamanders. Ohio J. Sci. 1, 12-15. adult newt (Triturus viridescens). Devi Biol. 17, 571-583. TASSAVA, R. A., JOHNSON-WINT, B. AND GROSS, J. (1986). GULATI, A. K., ZALEWSKI, A. A. AND REDDI, A. (1983). An Regenerate epithelium and skin glands of the adult newt react to immunofluorescent study of the distribution of fibronectin and the same monoclonal antibody. J. exp. Zool. 239, 229-240. laminin during limb regeneration in the adult newt. Devi Biol. TOOLE, B. P. (1981). Glycosaminoglycans in morphogenesis. In Cell 96, 355-365. Biology of Extracellular Matrix (ed. E. D. Hay), pp. 259-294. HYNES, R. O. (1981). Fibronectin and its relation to cellular New York: Plenum Press. structure and behavior. In Cell Biology of Extracellular Matrix TOOLE, B. P. AND GROSS, J. (1971). The extracellular matrix of the (ed. E. D. Hay), pp. 295-333. New York: Plenum Press. regenerating newt limb: Synthesis and removal of hyaluronate HYNES, R. O. (1985). Molecular biology of fibronectin. Ann. Rev. prior to differentiation. Devi Biol. 25, 57-77. Cell Biol. 1,67-90. TOWBIN, H., STAEHEUN, T. AND GORDON, J. (1979). INAGUMA, Y., KUSAKABE, M., MACKIE, E. J., PEARSON, C. A., Electrophoretic transfer of proteins from polyacrylamide gels to CHIQUET-EHRISMANN, R. AND SAKAKURE, T. (1988). Epithelial nitrocellulose sheets: procedure and some applications. Proc. induction of stromal tenascin in the mouse mammary gland: from natn. Acad. Sci. U.S.A. 76, 4350-4354. embryogenesis to carcinogenesis. Devi Biol. 128, 245-255. VAUGHAN, L., HUBER, S., CHIQUET, M. AND WINTERHALTER, K. H. ITEN, L. E. AND BRYANT, S. V. (1973). Forelimb regeneration from (1987). A major, six-armed glycoprotein from embryonic different levels of amputation in the newt, Notophthalmus cartilage. EMBO J. 6, 349-353. viridescens: Length, rate and stages. Wilhelm Roux Arch. EntwMech. Org. 173, 263-282. KELLY, D. J. AND TASSAVA, R. A. (1973). Cell division and (Accepted 20 January 1990)