Urodele amphibians are unique in being able to regenerate amputated limbs at both larval and adult stages of the life cycle. Important to limb regeneration is the formation of a blastema, a mass of undifferentiated mesenchyme cells covered by a multilayered epithelium (Wallace, 1981; Stocum, 1995; Tsonis, 1996). This epithelium is derived from skin epidermis within hours of limb amputation. Epidermal cells first migrate distally to cover the wound surface, forming a wound epithelium (WE; Thornton, 1957). They then proliferate (Hay and Fischman, 1961) to form the multilayered apical epithelial cap (AEC; Thornton, 1954; see also Globus et al., 1980). Blastema formation and growth, and thus successful regeneration, will not proceed without a functional AEC (reviewed in Wallace, 1981).
The AEC is a complex layered epithelium. The outermost cornified layer and the basal-most low columnar layer bound the inner layer(s) of cuboidal cells (Singer and Salpeter, 1961). Ultrastructural and immunocytochemical studies suggest a robust secretory activity, especially in the basal layer (Singer and Salpeter, 1961; Tassava et al., 1986; Goldhamer et al., 1989; Estrada et al., 1993). The AEC secretes factors needed by the underlying blastema mesenchyme, providing critical stimulatory signals for regeneration (reviewed in Stocum, 1995).
Classic experiments show that the AEC directs blastema outgrowth and is necessary for regeneration (Thornton, 1968). Blastemas deprived of the AEC do not form the complete proximodistal limb pattern, lacking distal structures (Stocum and Dearlove, 1972). If AEC formation is prevented by suturing a flap of skin over the wound surface or by inserting the freshly amputated end of the limb into the body cavity, regeneration will not proceed (Goss, 1956; Mescher, 1976; reviewed in Wallace, 1981). Thornton (1957) removed the AEC from the amputated limb stump on a daily basis; in Ambystoma tigrinum larvae, which are slow to recover the lost AEC, this daily removal prevents regeneration. Thornton (1960) also showed that the thickened AEC controls the direction of blastema outgrowth; when the central AEC is displaced to an eccentric position in an early blastema, the regenerate grows out at an angle. Addition of an ectopic AEC to the base of a blastema also directs limb outgrowth, resulting in an ectopic limb (Thornton and Thornton, 1965). Together with studies of the mitotic and labeling indices of blastemas, these results support the idea that an important function of the AEC is to direct blastema cell cycling and accumulation (Mescher, 1976; Globus et al., 1980).
The idea that the distal-most tissue stimulates limb outgrowth is not unique to the AEC and regenerating limbs. The distal-most tissue of the developing limbs of other vertebrates, the apical ectodermal ridge (AER), also stimulates limb outgrowth. Research shows that in the developing chick limb, an extra dorsal AER leads to extra dorsal digits (Goetinck, 1964; Saunders et al., 1976). The AER not only stimulates limb outgrowth, but it is also necessary for the complete proximodistal limb pattern. If the AER is removed early in limb development, distal outgrowth ceases and only the most proximal limb structures develop (Saunders, 1948).
However, correlations between the AEC and the AER are not exact. The AEC and the AER have obvious and important differences in size and shape (Tank et al., 1977). Whereas the AEC broadly covers the entire distal end of the limb stump, the AER is narrowly localized to the distal dorsoventral boundary of the limb bud. Also, the chick AER is made up of a single pseudostratified layer of ectodermal cells (Kelley and Fallon, 1983), whereas the salamander AEC is made up of true stratified layers, sometimes as many as 15 cell layers thick (Singer and Salpeter, 1961). Many studies support functions for the basal layer of the AEC that are distinct from the functions of the outer layers. Tenascin (TN), fibroblast growth factor receptor 1b (FGFR1b), fibronectin (FN), collagen XII (colXII), and connective tissue growth factor (CTGF) are all primarily expressed in the basal layer of the AEC (Onda et al., 1991; Poulin et al., 1993; Nace and Tassava, 1995; Wei et al., 1995; Cash et al., 1998). However, these studies focus mainly on comparisons between epithelial and mesenchymal cells rather than on comparisons within the AEC itself or on comparisons between the AEC and the AER.
In this study, we examine the AEC by using in situ hybridization and hematoxylin and eosin (H+E) staining. We investigate the basal layer of the AEC, i.e., the layer of cells closest to the distal mesenchyme, as the most active layer within the AEC. We also investigate whether the basal layer contains a subset of cells along the dorsoventral boundary similar to the AER of developing amniote limbs. The results indicate many biochemical and structural correlations between this basal layer and the amniote AER, which lead us to suggest that these two apical structures may be functionally homologous. We suggest that the basal AEC is the true functional layer of the AEC, the layer that promotes blastema cell cycling and controls the direction of limb outgrowth in regeneration.
Patterns of FN Expression: In Situ Hybridization
Regenerates displayed stage-specific patterns of FN expression as seen by antisense riboprobe signal. Whole-mounts revealed a stripe of FN expression along the dorsoventral boundary near the distal mesenchyme/epithelium (M/E) border in late-bud to palette stage blastemas (Fig. 1), whereas a more extensive cap-shaped expression was seen near the distal M/E border in dedifferentiation and early-bud stage blastemas (not shown). In all cases, stump epidermis was devoid of signal. No specific signal was seen in limbs processed with sense riboprobe (Fig. 1 and data not shown).
FN expression could not be specifically localized to mesenchyme or epithelium within the blastema by means of whole-mount observation alone. Therefore, sample limbs from the 12 limbs processed through whole-mount in situ hybridization were then embedded in OCT compound (Tissue-Tek, CA), frozen on dry ice, and cryosectioned at 30 microns. Limb orientations were carefully noted to further examine any differences in FN expression patterns between the anteroposterior (A/P) and dorsoventral (D/V) axes of the regeneration blastema. Observation of sections revealed that both the stripe and cap domains of expression were localized to the basal layer of the AEC. In A/P sections of late-bud stage regenerates (Fig. 2), almost all sections were devoid of signal except for a few sections through the very middle of the regenerate, presumably corresponding to the dorsoventral boundary. In these middle few sections, nearly the entire basal layer of the AEC showed strong reactivity. Adjacent sections were serially examined to confirm that expression was indeed found in a stripe-like pattern. In D/V sections of late-bud stage regenerates (Figs. 3, 4), every section through the distal tip showed expression in a cluster of 6–10 basal cells at the dorsoventral boundary of the AEC, and no expression further dorsal or ventral. In all late-bud stage regenerates, sparse mesenchymal reactivity was also apparent, but at a much lower magnitude than observed in the basal AEC. In sections of dedifferentiation and early-bud stage regenerates (Fig. 5), the entire basal layer of the AEC displayed strong reactivity, regardless of section angle. At these early stages, reactivity was evenly dispersed along both the A/P and D/V axes (not shown), whereas both the blastema cells and cells of the outer AEC appeared devoid of signal.
Further analysis of D/V sections of late-bud stage regenerates revealed a morphologic notch-like structure in the central-most sections (Fig. 4). The 6 central sections (of 13 consecutive sections showing probe reactivity) showed the cluster of 6–10 reactive cells folded above the basal lamina in a “V” (outer middle sections) or tight “U” (inner middle sections) shape. This notch represents the cross-section of an A/P structural groove, which has been previously described in the AER of avian and reptilian developing limbs (Tomasek et al., 1982; reviewed in Todt and Fallon, 1984, 1986).
To determine whether the cluster of FN-expressing basal AEC cells was visible without FN in situ hybridization as a marker, sections of late-bud regenerates were stained with H+E (Fig. 6). Limbs sectioned along the D/V axis and stained with H+E occasionally showed a cluster of elongated columnar cells in the basal AEC at the dorsoventral boundary. Later stage regenerates (early palette) and thinner, faster-growing limbs (hindlimbs) seemed to have more elongated cells throughout the regenerate than earlier stage and thicker limbs. The cluster of elongated basal cells was only seen in the central-most limb sections and was often quite discrete, as was the underlying notch (compare Fig. 6A,B). The notch was not deep or pronounced (as in Fig. 4A) in any of these sections, and its location and visibility varied with section location and angle. Because the groove runs along the A/P axis, no notch was visible in A/P sections. A/P sectioning also proscribed consistent identification of elongated basal cells (data not shown). These differences in basal cell cluster and notch visibility based on section angle, section location, and regenerate stage and size perhaps explains why these structures were not previously noted along the dorsoventral boundary in the basal AEC.
Additionally, H+E staining provided insight into other distal structures and their relation to the AEC. Basement membranes stain very darkly in H+E and were found to always be absent under the AEC yet present under the stump epidermis (not shown). Leydig (gland) cells of urodele epidermis are also clearly visible as large white cells that are consistently absent from the basal layer of the AEC (Fig. 6A).
In this study, we present an analysis of the compartments of the apical epithelial cap of regenerating limbs, including the differential expression of FN message within the AEC. The results indicate that the basal layer of the AEC may have a role distinct from the other layers based on its intense expression of FN message and distinct structure. We suggest that the role of the basal layer may be homologous to the role of the AER in the developing amniote limb. We also suggest further ultrastructural and functional comparisons to test this hypothesis. Our studies of the compartments of the AEC also lead us to suggest a more consistent and unifying nomenclature.
Through in situ hybridization, we have identified the basal layer as a specialized portion of the AEC. FN message is expressed at a higher level in the basal layer of the AEC than in other areas of the regenerating limb (see also Nace and Tassava, 1995). FN message is not detectable in the basal layer of the adjacent stump epidermis. The high FN expression in the basal AEC cells extends across the entire amputated limb tip a few days after amputation, but then becomes restricted to a stripe of expression along the dorsoventral boundary as regeneration proceeds. In addition, the basal layer appears to fold along this dorsoventral boundary to form a structural groove, seen as a notch in cross-sections. H+E staining also shows elongated basal cells at the dorsoventral boundary, and the lack of gland cells throughout the basal layer. Together, these observations suggest that in regenerating limbs the basal layer of the AEC has unique structures and activities, many of which appear similar to the AER of developing amniote limbs.
Discussion of Nomenclature
The apical epithelial cap is a complex and variable structure. It was described by Thornton (1956b) as “richly innervated tip (epithelium which) formed a thickened cap under which the blastema cells aggregated.” Unfortunately, such descriptions of the AEC have left some confusion as to the difference between the AEC and the wound epithelium (WE) and the terms are now often used interchangeably (e.g., Singer and Inoue, 1964; Wallace, 1981; Onda and Tassava, 1991). Here, we suggest the term “wound epithelium” to mean the 1–3 layers of epithelial cells that initially migrate over the wound (Thornton, 1957; Singer and Salpeter, 1961). Thornton (1954) suggests that these cells that initially cover the wound surface are only a precursor to the AEC, and merely represent a part of the early wound-healing process. Their purpose is to provide a covering for open wounds — a “wound epithelium” (Thornton, 1957).
In contrast to wound epithelium, we suggest that the term “AEC” be restricted to the thickened epithelium that accompanies those stages of regeneration when mesenchymal cells are dedifferentiating, accumulating, and dividing. Thus, the AEC is usually four or more layers thick and may become mildly mitotically active, unlike the migrating wound epithelium (Thornton, 1957; Hay and Fischman, 1961). The AEC is maintained until its eventual proximal to distal transformation into skin epidermis during late stages of regeneration (Singer, 1949; Thornton, 1954, 1956a Salpeter and Singer, 1960). Thus, in normal regenerating limbs the use of the terms WE or AEC can be largely determined by the time or stage after amputation.
As additional notes on terminology, we use here the term “epithelial” cap as opposed to “epidermal” cap because the AEC lacks a basement membrane and the subjacent dermal components characteristic of an epidermis (Thornton, 1954; Ruben and Frothingham, 1958; Salpeter and Singer, 1960; Stocum, 1985). In addition, Thornton suggests that not all parts of the regenerate epithelium are equivalent, as evidenced by his experiments with eccentric caps and with caps grafted to the sides of blastemas (Thornton, 1960; Thornton and Steen, 1962; Thornton and Thornton, 1965). In such cases, we suggest the terms “central AEC” for that portion of the AEC which directs blastema outgrowth, and “peripheral AEC” for the thinner sides of the AEC (Onda and Tassava, 1991). It is our hope that these definitions will aid in discussing structural, functional, and stage-specific differences in the WE, AEC, and skin epidermis of regenerating limbs.
Differences Between AEC Compartments
Consider now the different compartments within the multilayered AEC. The AEC consists of three structurally different cell types: outer keratinized, middle cuboidal, and basal low columnar. Locational and simple structural differences between the three types of layers in the AEC suggest that they may have different functions. The outermost keratinized layer is likely protective. The middle cuboidal layer may be a transition cell type proceeding from functional basal cells to keratinizing outer cells. The low columnar basal layer may secrete factors proximally to influence outgrowth of the underlying mesenchyme (Mescher, 1976). When Leydig (gland) cells replace the basal layer, the AEC appears nonfunctional (Thornton, 1958; Maden and Wallace, 1976). We suggest that it may be only the basal layer which mediates AEC function and dictates the geometric angle of limb outgrowth (Thornton, 1960; Thornton and Steen, 1962). We discuss here the unique structures and activities, as well as possible roles, of this basal layer.
The location and activity of the basal layer make it a unique compartment of the AEC. The cells of the basal layer are the closest epithelial cells to the distal mesenchyme of the blastema, and studies suggest that these cells have robust secretory activity (Singer and Salpeter, 1961; Goldhamer et al., 1989). Here, we show that FN message in the AEC is expressed only in the cells of the basal layer, and previous work showed FN protein strongly immunolocalized beneath the basal layer (Nace and Tassava, 1995); thus, FN protein appears to be secreted into the basal lamina, mesenchyme, or both, from these cells. It is likely that growth factors are also secreted by the basal cells (reviewed in Stocum, 1995). This growth factor signaling hypothesis corresponds well with previous studies of AEC activity. Thornton (1960) and Thornton and Steen (1962) suggest a directive role for the AEC, meaning that the location of the AEC determines the geometric direction/angle of limb outgrowth. A possible mechanism of action for this directive role is the secretion of molecules from the basal layer. These molecules then promote proliferation and distal accumulation of the blastema mesenchyme cells (Mescher, 1976; Globus et al., 1980).
The production of FN solely in the basal AEC supports the idea that the basal layer has functions beyond those of the outer layers. FN is an important building block of the extracellular matrix, and it is expressed strongly in wound epithelia. In vitro studies have identified many potential roles for FN, including cell proliferation, cell adhesion, cell migration, and cell differentiation (Oberley et al., 1983; Turner et al., 1983; Mooradian et al., 1993; Crisa et al., 1996; Salomon et al., 1997). Additional studies have shown that FN has critical effects on cell spreading, migration, and proliferation in wound tissues, including increasing the rate of wound healing (Nishida et al., 1983; Caron et al., 1985; Donaldson et al., 1987; Watanabe et al., 1991; Garat et al., 1996). FN levels, thus, are tightly regulated during wound healing (Nishida et al., 1982; Phan et al., 1989; Ruoslahti, 1991; Pawar et al., 1995; Asari et al., 1996; Goke et al., 1996). Just as in the AEC, the only epithelial layer to up-regulate and express FN in deeply wounded rat corneas is the basal layer (Nickeleit et al., 1996). Because there is an obvious need for cells to replace those that have been wounded or lost in limb regeneration and corneal healing, some combination of cell proliferation, adhesion, migration, and spreading is also necessary to provide the needed cells. Thus, FN is up-regulated in the basal epithelia of these wound tissues where its proposed functions would be most needed, i.e., in areas of cell recruitment and growth.
Comparing the AEC and the AER
Other areas of tissue growth and remodeling, such as the developing limb bud, also accumulate FN. FN protein localization appears very similar in regenerating and developing limbs, especially in the distal tip. FN protein is heavily localized immediately subjacent to the AEC in regenerating limbs (Nace and Tassava, 1995) and immediately subjacent to the AER in developing limbs (Tomasek et al., 1982). In both cases, FN accumulates in the basal lamina (Kosher et al., 1982; Tomasek et al., 1982; Nace and Tassava, 1995; Kanazawa et al., 1998). In late stages, both the AER and the AEC have a groove in the basal layer where FN is most concentrated, likely indicating the definitive tip of limb growth (Tomasek et al., 1982; Todt and Fallon, 1984, 1986). Gehris et al. (1996) report that FN message is indeed expressed in the AER. Further, FN immunolocalization supports this idea, showing granular reactivity within AER cells that appear to be exporting FN to the basal lamina (Kanazawa et al., 1996). These data suggest that the AER is secretory in nature as well, an idea already well accepted (e.g., Niswander et al., 1994). Thus, both the basal AEC and the AER appear to produce FN and secrete it to the underlying lamina.
Despite the similarities in activity between the basal AEC and the AER, there are other important differences. The typical AER structure does not appear concurrently with the initial limb bud but instead appears after outgrowth (stages 18–20 for chick; Todt and Fallon, 1984) and persists until the formation of digits. The AEC persists until the formation of digits as well, but, in contrast to the AER, the AEC seems to appear earlier, i.e., before outgrowth (Thornton, 1958). This early AEC formation can also be seen in its characteristic expression of matrix genes as early as 1–3 days postamputation (Onda et al., 1991; Nace and Tassava, 1995; Wei et al., 1995).
Perhaps this timing discrepancy can be best understood by structural comparisons of the two apical structures. The AER is the absolute distal-most structure in the developing limb. It is found only at the dorsoventral boundary in the shape of a long semi-circle from anterior to posterior (Todt and Fallon, 1984). The AEC, on the other hand, is initially a flat covering of the amputation plane. It then thickens in the radial center of the limb and moves distal-ward as if pushed outward/upward like a circus tent by the accumulating blastema cells beneath. At the late-bud stage (Stocum, 1979) of blastema development, however, the blastema begins to flatten dorsoventrally (Tank et al., 1976) to form a paddle shape (palette stage) and the AEC flattens with it. Our FN in situ hybridization results here show that at these later stages of regeneration, FN is expressed along the flattening tip, in a semi-circular stripe from anterior to posterior, in cells forming an internal cryptic ridge rather than an external morphologic ridge. Thus, after the underlying blastema has formed and begins to resemble more closely the developing limb bud, the basal AEC begins to resemble more closely the AER, linking the timing and structures of these two apical structures.
The present study buttresses previous molecular studies that indicate correlations between the AER and the AEC (Mullen et al., 1996; Carlson et al., 1998). However, FN mRNA localization to the basal layer of the AEC suggests a potential refinement of this correlation, i.e., that perhaps only the basal compartment of the AEC is functionally homologous to the AER. Additional studies will be needed, however, to determine whether any refinement is necessary. Scanning electron microscopy reveals gap junctions as indicator structures of the AER (Fallon and Kelley, 1977); however, electron microscopy studies in regenerating limbs have not yet focused on gap junctions in the basal AEC (Bryant et al., 1971; Vethamany-Globus et al., 1993). In addition, FN protein localization is well correlated between the two apical structures; however, FN in situ hybridization has not yet focused on FN expression levels around the groove in the AER.
Nevertheless, without further functional analysis, these studies cannot extend the correlation of these structures to indicate identical functions. The key functional molecules produced by the AER are thought to be fibroblast growth factors 4 and 8 (FGF4, 8), due to the exclusive expression in the AER of their respective mRNAs and the ability of these growth factors to replace the AER in a developing limb (Niswander and Martin, 1992; Niswander et al., 1993; Crossley et al., 1996; Vogel et al., 1996). Christen and Slack (1997) have shown that FGF8 expression appears identical in developing and regenerating Xenopus laevis limbs and correlates with successful regeneration. Han and Kim (1998) reported FGF8 expression by the axolotl blastema but not cellular localization. Work is currently under way in our laboratory to investigate these key functional members of the FGF family with regard to their expression and functional significance in the basal layer of the AEC.
Axolotl larvae were obtained from the Indiana University Axolotl Colony, fed beef liver 3 times/week and raised to 4.0 cm snout-tail tip length at 22°C. Axolotls were separated into individual dishes to minimize predation and limb chewing. Axolotls were anesthetized in MS222, and undamaged limbs were amputated through the distal zeugopod (humerus or femur). Limbs/regenerates were sampled at initial dedifferentiation, early-bud, late-bud, and palette (early redifferentiation) stages and processed for in situ hybridization (staging after Stocum, 1979). Additional limbs were amputated through the proximal stylopod (radius/ulna or tibia/fibula), sampled at early-bud and late-bud stages, and processed for H+E staining.
In Situ Hybridization
An adaptation of the procedure of Gardiner et al. (1995) was used for whole-mount in situ hybridization to probe for fibronectin (FN) message. Sense and antisense digoxigenin-labeled probes were generated in T3 and T7 in vitro transcription reactions (Boehringer Mannheim, IN). Probes were generated from a 471-bp fragment of a urodele FN cDNA (Genbank accession S76886; Nace and Tassava, 1995) cloned into the vector pBluescript SK+ (Stratagene, CA). The cloned region of FN contains three conserved FN type I domains and part of a FN type II domain (Nace and Tassava, 1995).
Twelve limbs/regenerates were prepared for in situ hybridization by fixation overnight in 3.7% formaldehyde followed by 100% methanol washes. Limbs/regenerates were dissected, rehydrated in PTw (1X phosphate-buffered saline, 0.1% Tween-20), and treated with 20 μg/ml proteinase K for either 5 or 15 min. Limbs/regenerates were then acetylated by using 0.1 M triethanolamine and acetic anhydride, washed with PTw, refixed in 3.7% formaldehyde, and washed again with PTw. Limbs/regenerates were prehybridized in hybridization solution (50% formamide, 5× standard saline citrate (SSC), 1 mg/ml yeast RNA, 100 μg/ml heparin, 1× Denhardt's, 0.1% Tween-20, 0.1% CHAPS, and 5 mM ethylenediaminetetraacetic acid) overnight at 50°C. Probes were heated to 80°C for 2 min and hybridization was carried out at 50°C for 3 days. Limbs/regenerates were then washed in 2× SSC and 0.2× SSC at 55°C, in MAB (100 mM maleic acid, 150 mM NaCl, pH 7.5) and MAB-B (MAB, 2 mg/ml bovine serum albumin) at room temperature, and in blocking solution (MAB-B, 20% sheep serum) at 4°C. Digoxigenin antibody (Boehringer Mannheim) was diluted in blocking solution and applied to limbs/regenerates overnight at 4°C. Limbs/regenerates were then rinsed for 5 hr in 10 changes of MAB and the NBT/BCIP color reaction was performed for 1 hr in the dark at room temperature (Boehringer Mannheim). Only a short reaction time was necessary to observe the cells most actively expressing FN message. The reaction was then terminated by transferring the limbs to phosphate buffered formalin. Limbs/regenerates were observed and photographed as whole-mounts, embedded in OCT compound, or both, snap frozen on dry ice, and cryosectioned at 30 microns. Sections were air-dried and mounted in glycerol for observation and photography.
Eight limbs/regenerates were prepared for H+E staining by fixation overnight in Bouin's fixative, processed, and embedded in paraffin. Embedded tissues were sectioned at 10–12 microns, air-dried, deparaffinized, stained in Delafield's H+E, cleared in toluene, and mounted with picolyte for observation and photography under brightfield and/or fluorescence microscopy.
The authors thank Dr. Eric V. Yang for assistance with the whole mount in situ hybridization procedure as adapted from that of Drs. David Gardiner and Susan V. Bryant of UC Irvine (Gardiner et al., 1995). We also thank Drs. Essam Moussad, Kenneth P. Klatt, and Anthony L. Mescher for their helpful critiques of the manuscript.