tenascin localization in skin wounds of the adult newt notophthalmus viridescens

THE ANATOMICAL RECORD 230:451-459 (1991)

Tenascin Localization in Skin Wounds of the Adult Newt Nofophthalmus viridescens

DONALD J. DONALDSON, JAMES T. MAHAN, HUI YANG, AND KATHRYN L. CROSSIN

Department of Anatomy and Neurobiology, University of Tennessee Center for the Health Sciences, Memphis, Tennessee (D.J.D., J.T.M., Y.H.); The Rockefeller University,

New York, New York 10021 (K.L.C.)

ABSTRACT Earlier studies have shown that the extracellular matrix (ECM) protein tenascin (TN) is present between uninjured epidermal cells of urodele appendages, but is absent from most of the mesenchymally derived ECM. Follow- ing appendage amputation, this distribution is reversed. TN is lost from the epi- dermis and appears in the ECM of the stump and the regeneration blastema. In the present study, monoclonal and polyclonal antibodies to TN were used to localize this protein immunohistochemically in limbs of the adult urodele Notophthalmus viridescens a t various stages following skin removal with or without damage to underlying muscle to determine 1) if the loss of TN by the epidermis and its gain by mesenchymal tissues occurs in wounds that do not require regulation by epi- genetic mechanisms, and 2) if TN is present in the provisional wound matrix beneath migrating epidermal cells. In addition, skin explants were cultured on TN-coated dishes to learn if TN possesses active sites that can support epidermal cell migration. The results indicate that simple wounding leads to the same TN patterns as occurs following limb amputation. Tenascin loss from the epidermis could be seen as early as 6 hr after wounding, a time during which migrating epidermal cells are moving over the wound bed. During this period, there was no evidence of TN in the provisional wound matrix. In contrast to collagen, which supports considerable epidermal cell migration from skin explants, TN allowed no more migration than did the inactive protein, myoglobin.

The ability of epidermal cells to engage in the coor- dinated movements required for wound closure is de- pendent on the presence of appropriate macromolecules in the provisional extracellular wound matrix. Known components of this matrix include fibrinogen and fibro- nectin (Clark et al., 1982), proteins that are able to mediate newt epidermal cell migration through inter- action with epidermal receptors capable of recognizing the amino acid attachment sequence, Arg-Gly-Asp- SER (RGDS) (Donaldson et al., 1987). Newt epidermal cells are also able to migrate on the ECM proteins, collagen, and vitronectin through the same or similar receptors as those involved in migration over fibrino- gen and fibronectin (Donaldson et al., 1987). Migration does not occur, however, on a variety of other proteins such as casein, myoglobin, fetuin, and BSA (Donaldson and Mahan, 1984). While there is evidence that newt epidermal cells do not require the RGD signal to mi- grate (Donaldson et al., 1988), the presence of such a sequence in a given protein suggests that it might have epidermal migration-supporting capacity. Tenascin (TN), a recently discovered extracellular matrix pro- tein, falls into this category.

Known variously as myotendinous antigen (Chiquet and Fambrough, 1984a,b), glioma mesenchymal extra- cellular matrix protein (Bourdon et al., 1985), cytotac- tin (Grumet et al., 19851, J1 (Kruse et al., 1985), and hexabrachion (Erickson and Iglesias, 1984), tenascin

(c) 1991 WILEY-LISS, INC

(Chiquet-Ehrismann et al., 1986) is a high molecular weight glycoprotein with six apparently identical RGD-containing subunits, each of which forms one of its six arms (see review by Erickson and Bourdon, 1989). The expression of TN in developing tissues is so spatially and temporally regulated that i t is believed to play an important role in morphogenetic events (Tan et al., 1987). Tenascin is common in mesenchymal tumors and carcinomas and may be present even when it is absent from the normal tissue of origin (Mackie et al., 1987). Though i t is largely absent from unwounded adult rat skin, TN rapidly appears in the dermis adja- cent to a wound and in the granulation tissue that develops later (Mackie et al., 1988). The fact that form- ing wound epithelium was always underlain by TN in these latter studies prompted the authors to suggest that TN might somehow facilitate epidermal cell mi- gration.

Two studies have described the immunolocalization of TN in normal and regenerating urodele appendages. In the mesenchymal tissues of normal adult newt limbs

Received September 25, 1990; accepted February 4, 1991. Address reprint requests to Donald J. Donaldson, Department of

Anatomy and Neurobiology, University of Tennessee Center for the Health Sciences, Memphis, TN 38163.

452 D.J. DONALI

(Onda et al., 1990) and the normal tail of larval Pleu- rodeles waltz (Arsanto et al., 1990), TN was restricted to discrete locations such as tendons, ligaments, myo- tendinous junctions, periosteum, and perichondrium. Most of the extracellular matrix (ECM), however, was TN-negative. Surprisingly, TN was also found in the epidermis where it appeared to be in the intercellular space between keratinocytes. Within a few days follow- ing amputation of either appendage, there was wide- spread acquisition of TN by previously negative stump connective tissue. The ECM of regeneration blastemas was likewise strongly TN-positive. Wound epithelium, however, showed a great reduction in TN such that on newt limbs TN was absent a t 1 day post-amputation (the earliest stage examined) and thereafter was only marginally present until digit differentiation began. On tails, wound epithelium was TN-negative from day 4 (the earliest stage for which data was presented) through day 14.

In the present study, we have sought to determine if the apparent loss of TN in newt wound epithelium and the acquisition of TN by mesenchymal tissue can occur in response to wounds that do not involve amputation. In so doing, we were particularly interested in localiz- ing TN during the migratory stage of wound closure [a stage not examined by either Onda et al. (1990) or Ar- santo et al. (199011 to learn whether this protein is a component of the provisional wound matrix, where it might function as a substrate for translocating epider- mal cells. We also tested purified TN directly for its ability to support newt epidermal cell migration.

MATERIALS AND METHODS Animals

Adult male newts (Notophthalmus uiridescens) were purchased from the Connecticut Valley Biological Sup- ply Co., Southampton, MA. Details of animal mainte- nance have been described previously (Donaldson and Mahan, 1983).

Experimental Wounds Under general anesthesia produced by whole animal

immersion in 0.15% 3-aminobenzoic acid ethyl ester (MS 222, Sigma Chemical Co., St. Louis, MO), a full thickness rectangle (1.5 x 3 mm) of skin was removed from the dorsal surface of each hind limb between the knee and ankle (type I wounds). Usually, the superfi- cial fibers of the underlying skeletal muscle were then minced with iridectomy scissors (type I1 wounds). An- imals were placed in tenth strength Holtfreter’s solu- tion (an amphibian saline) containing streptomycin sesquisulfate (Sigma) at a final concentration of 5 mg/L to allow healing. For immunocytochemistry, sections from at least four animals were examined for each con- dition as indicated in the figure legends.

Primary Antibodies Tenascin was localized through the use of MT1, a

monoclonal IgM kindly provided by Dr. Roy Tassava, Ohio State University. Made against homogenates of newt limb regeneration blastemas, MT1 appears to be directed against newt TN by a variety of criteria. Thus, a polyclonal antibody against chick TN competes effec- tively for MT1 binding sites in sections of regeneration blastemas, the two antibodies both stain a protein with

ISON ET AL.

the molecular weight of intact TN in immunoblots of blastemal extracts, and both stain this same protein in MT1 immunoprecipitates from blastemal extracts (Onda et al., 1990). As a control for MT1 staining, we utilized MC-480, an IgM monoclonal directed against the SSEA-1 carbohydrate epitope on glycolipids and glycoproteins in mouse embryos and embryonal carci- nomas (obtained from the Developmental Studies Hy- bridoma Bank a t Johns Hopkins School of Medicine).

In addition to MT1, two different polyclonal prepa- rations were used to localize TN in normal skin and 24 hr wound epithelium. The one used in Figure 6 was a rabbit anti-chick brain cytotactin (TN) serum previ- ously described and characterized in Hoffman et al. (1988). Identical results were obtained with a guinea pig anti-chick TN serum (0-45% ammonium sulfate cut dialyzed against PBS), prepared by the same method as the rabbit anti-TN described in Chiquet- Ehrismann et al. (1986) (a kind gift from Dr. Eleanor Mackie). Nonimmune rabbit serum and nonimmune guinea pig serum (0-45% ammonium sulfate cut dia- lyzed against PBS) were used as controls.

To prepare a polyclonal antibody against newt fibro- nectin (FN), FN purified from newt plasma by affinity chromatography on gelatin-agarose (Sigma) was sub- jected to preparative SDS polyacrylamide gel electro- phoresis under reducing conditions. The region of the gel containing FN (220 kD) was excised, emulsified in adjuvant, and used for rabbit immunization. Anti-newt FN IgG and preimmune IgG were purified by protein-A chromatography. Both anti-newt FN IgG and preim- mune IgG were absorbed with FN-depleted newt plasma to increase specificity. After absorption, anti- FN IgG stained two bands in immunoblots of newt plasma, both of which comigrated with corresponding bands in human FN standards.

lmmunocytochemistry For TN localization, cryostat sections of newt tissue

were incubated in MT1 (approx. 90 Fg of immunoglob- uliniml of PBS), a t room temperature (rm temp) for 1 hr followed by three washes in 0.05% Triton/PBS. After an additional PBS wash, the sections were exposed to anti-mouse immunoglobulin conjugated to Texas Red (Vector Labs, Burlingame, CA) in PBS for 1 hr a t rm temp. Sections were then washed again and cover- slipped with glycerol/PBS (3:l) containing 2.5% 1,4- diazabicyclo [2.2.2] octane (Sigma) to retard bleaching.

Polyclonal antisera against TN were used at 1:lO and 1 5 0 dilutions in PBS containing 1% BSA for 1 hr a t rm temp. Following three PBS washes, the sections were incubated for 1 hr at rm temp in either goat anti- guinea pig or anti-rabbit IgG, both of which were con- jugated to Texas Red (Vector). Washing and coverslip- ping were as described above.

To localize FN, cryostat sections or whole mounts of wounds that had not yet formed a wound epithelium were treated overnight a t 4°C with polyclonal anti-FN IgG that had been absorbed with FN-depleted plasma. Bound antibodies were visualized by exposure of the treated tissue sequentially to avidin- and biotin- blocking solutions, biotin-labeled guinea pig anti- rabbit IgG, and avidin Rhodamine (all from Vector). Whole mounts were cryosectioned and all sections were washed and coverslipped as above.

TENASCIN AND SKIN WOUNDS 453

Skin Explants Cultured on TN Tissue culture dishes were coated with the indicated

amounts of either bovine type I collagen (obtained from Dr. Mustafa Dabbous, Dept. of Biochemistry, Univer- sity of Tennessee, Memphis, TN), equine myoglobin (Sigma), or human TN (Telios Pharmaceuticals, San Diego, CA). Ten microliter aliquots of test proteins were placed in 50 mm2 circles (six per dish) and al- lowed to dry overnight at 23°C. Before use, the dishes were washed four times with distilled water. The col- lagen was made up in 0.1 M acetic acid; myoglobin and TN were in PBS. Pieces of full-thickness newt skin, as described earlier under Experimental Wounds, were then explanted onto the coated spots so that for each explant placed on myoglobin or TN a control explant from the contralateral limb was placed on collagen. Five milliliters of 60% CEM culture medium (B&B/ Scott Labs., Fiskeville, RI) was then added to each dish. After 16 h r incubation a t 23"C, the explants were fixed in 10% formalin. The amount of migration in each ex- plant was determined planimetrically as described pre- viously (Donaldson et al., 1987). To compare the migra- tory performance on each experimental substrate to the performance on collagen, the following formula was used:

planimeter value for an individual explant group mean for the appropriate contralateral collagen controls

x 100.

These calculated values were then used to generate the means shown in Figure 7.

RESULTS TN Localization in Normal Limbs Using the

Monoclonal, MTl Newt skin consists of a stratified squamous epithe-

lium four or five cells thick resting on a connective tissue dermis that contains large oval mucous glands and numerous melanocytes. In the dorsal part of the hind limbs (where these experiments were conducted), the dermis, in turn, rests on the limb musculature (in- set in Fig. 1). In samples from unwounded limbs (skin plus a small amount of superficial muscle), MT1 reac- tivity was limited to tendons, portions of the interface between glands and the adjacent tissue, and, most sig- nificantly for this study, the epidermis. Within the epi- dermis, MT1 immunoreactivity was punctate and ap- peared to be localized in the intercellular space where it surrounded the apical ends of basal cells and often completely enclosed cells in the second and third lay- ers. In some areas, positive fluorescence was also seen along the dermo-epidermal junction (Fig. 1). Dermal connective tissue, gland cells, and underlying skeletal muscle were negative. No fluorescence was detected in control sections incubated with an equivalent concen- tration of MC-480 (a monoclonal antibody that, like MT1, is of the IgM isotype).

TN Localization in Wounded Limbs Using MT1 Within 6 h r after limbs were inflicted with type I1

wounds (the earliest time point sampled), TN immu- noreactivity in the epidermis immediately adjacent to the wound was much reduced. This can be seen in Fig- ure 2, which shows a segment of epidermis (approxi-

mately 1 mm from the wound) in which only scattered particulate fluorescence remains. No TN immunofluo- rescence was detected in the forming wound epithelium (Fig. 3). Nor did we see any evidence of TN on the wound bed (arrow in Fig. 3). This is in contrast to fi- bronectin (FN) staining which could be seen on wound beds when a polyclonal anti-FN was used to treat ei- ther tissue sections or dissected wounds which were sectioned after antibody exposure. This is shown in the inset in Figure 3 where anti-FN treatment prior to sectioning has produced a continuous immunoreactive band on the wound bed. Absorption of anti-FN with purified newt FN abolished the staining. Presumably, if FN is available to antibodies, i t is also available to migrating cells. To our knowledge, this is the first di- rect evidence in any species that wound-associated FN is actually exposed on the wound bed.

After 24 hr, type I1 wounds were completely closed by a multilayered wound epithelium. Under the wound epithelium, injured muscle fibers were beginning to break down (Fig. 4A). Tenascin appeared to be absent from the epidermis bordering the wound (Fig. 4B) and from the wound epithelium (Fig. 4C). Some TN immu- nolabeling was observed in the deeper tissues but this was limited to lengths of tendon, which are positive even before wounding, and infrequent accumulations of what appeared to be a vacuolated exudate (Fig. 4C).

In the ensuing days and weeks, injured muscle fibers degenerated and were replaced by a highly cellular wound matrix that stained heavily for TN (Fig. 5). No labeling of this matrix was seen when the control monoclonal, MC-480, was used in place of MT1. Even after 3 weeks, there was little if any TN immunoreac- tivity in either the epidermis immediately bordering the wound (Fig. 5B) or in the wound epithelium (Fig. 5C). Since the experiment was not carried beyond 3 weeks, we do not know when normal epidermal TN reactivity reappears.

Tenascin also appeared under the wound epithelium in type I wounds where the skin was removed without damage to underlying muscle. Surprisingly, in addi- tion to finding TN in the reforming dermis, we also found it around the more superficial muscle fibers un- der the wound (not shown). The epidermal pattern was identical to that described above for type I1 wounds.

TN Localization in Wounded Limbs Using Polyclonal Anti- TN

To determine if the loss of immunoreactivity de- tected with MT1 was limited to the MT1 epitope, we stained 24 hr wound epithelium with two different polyclonal antisera against chick TN. Each antiserum produced the result shown in Figure 6 where the 24 hr wound epithelium (we) is essentially devoid of stain while the epidermis from the contralateral unwounded limb (ne) shows the typical TN staining pattern of nor- mal skin. Thus, the reduction in TN immunolabeling cannot be explained by the loss or masking of a single epitope, but probably represents an effect involving the entire TN molecule.

TN as a Migration Substrate A major goal of this study was to determine if TN

might play a role in wound closure by acting as a sub- strate for epidermal cell migration. Our immunohisto-

454 D.J. DONALDSON ET AL.

Figs. 1-3

TENASCIN AND SKIN WOUNDS 455

chemical results showing loss of TN immunoreactivity in migrating epidermis could mean that TN is not present in the wound environment during the time when a wound epithelium is forming. If this is true, then epidermal cells clearly do not accomplish wound closure by migrating on this protein. It may be, how- ever, that the amount of TN present in the wound en- vironment had simply fallen below our level of detec- tion. We therefore decided to test TN directly for its ability to support migration. Pieces of newt skin were explanted onto plastic dishes coated with collagen (a positive control), myoglobin (a negative control), or TN and were then incubated in culture medium for 16 hr. In this system, proteins like collagen support the for- mation of a considerable halo of migrated epidermal cells around the explant in the time allowed. Unlike the robust response to collagen, migration on TN was not significantly different from migration on the neg- ative control, myoglobin (Fig. 7). This failure of TN to support migration in a direct trial and the immunohis- tochemical evidence discussed above is consistent with the idea that TN is not a migration-supporting compo- nent of the provisional wound matrix during epidermal closure.

DISCUSSION Tenascin is unique among extracellular matrix mol-

ecules in that i t possesses both a cell binding site and an anti-adhesive region for certain cells (Spring et al., 1989). The adhesive site resides in the distal region of each of the six arms that surround the central core of the molecule (Friedlander e t al., 1988; Spring et al., 1989). Most cells that adhere to TN either do not spread (Chiquet-Ehrismann et al., 1988) or spread less com- pletely on i t than they do on substrates such as fibro- nectin (Bourdon and Ruoslahti, 1989). However, quail neural crest cells not only adhere to TN but also effec- tively migrate on it (Halfter et al., 1989). These obser- vations coupled with those of Onda et al. (1990) and

(Figs. 1-3) Fig. 1. Cryostat section of newt skin and underlying muscle from an unwounded limb stained with a monoclonal antibody to TN. The inset shows a section through a similar region stained with hematoxylin and eosin (H & E). Tenascin is found only in the intercellular spaces between epidermal cells, in certain regions of the dermo-epidermal junction, and in portions of the interface between glands and dermal connective tissue. g = glands, m = muscle. The arrows point to the dermo-epidermal junction. Magnification of main panel, 162 x ; inset, 50 x .

Fig. 2. Cryostat section through uninjured skin and underlying muscle from a region adjacent to a 6 hr type I1 wound stained with a monoclonal antibody to TN. Note how TN staining in the epidermis has been reduced from the extensive pattern in Figure 1 to scattered dots. m : muscle. The arrows point to the dermo-epidermal junction. 162 x .

Fig. 3. Cryostat section through the leading edge of the forming wound epithelium in a 6 hr type I1 wound stained with a monoclonal antibody to TN. The rectangle in the upper inset ( a section stained with H & E) shows the region pictured in the main panel. The dotted line in the main panel outlines the advancing wound epithelium. Note that TN staining is absent from both the migrating epidermis and the wound bed (arrow). The lower inset shows a thin band of fibronectin on the wound bed revealed by exposure of a fresh type I wound to fibronectin antibodies before sectioning. m = muscle. Mag- nification of main panel, 162 x ; upper inset, 45 x , lower inset, 78 x .

Arsanto et al. (1990) in which TN was found in unin- jured urodele epidermis, and the findings of Mackie e t al. (1988), in which the appearance of TN in wounded rat skin correlated both temporally and spatially with migration of epidermal cells during wound closure, suggested that TN might play an important role in formation of a wound epithelium in urodeles. One way it could do this would be by functioning directly as a substrate for keratinocyte migration. Information on the likelihood of this possibility was obtained by im- munohistochemical localization of TN in normal and wounded newt limbs.

In unwounded newt skin, TN immunoreactivity, as determined with the MT1 monoclonal, is found prima- rily as punctate deposits that appear to be between epidermal cells. Following removal of a small piece of skin with or without damage to underlying muscle, there is a relatively rapid loss of TN staining, suggest- ing that the epidermal cells composing the wound ep- ithelium contain little if any TN. Nor is there any de- tectable TN on the surface over which the wound epithelium migrates, suggesting that TN is not a major component of the provisional wound matrix.

The monoclonal antibody that we used in this study recognizes TN in its native but not in its reduced form (Onda et al., 1990), which suggests that the epitope recognized is in the disulfide-rich central region of the molecule, where the six arms meet (Erickson and Bour- don, 1989). Had we limited our localization experi- ments to the use of MT1, i t could be argued that i t was not the entire TN molecule that was lost from migrat- ing epidermal cells, but only a portion containing the MT1 epitope. Or the epitope in question may have somehow become masked. The fact that wound epithe- lial reactivity to two different polyclonal anti-TN sera, which presumably are directed against multiple epitopes, was also lost weakens the above arguments. We think that loss of immunoreactivity is most likely due to proteolytic loss of either a large immunogenic region or loss of the entire TN molecule rather than masking of all available immunogenic epitopes. The localization data therefore suggest that TN does not function as a keratinocyte migration substrate.

The apparent lack of TN along the migratory path of keratinocytes is in distinct contrast to fibronectin, which forms a thin coating over the wound bed and which, by whole mount staining with anti-fibronectin, appears to be available to epidermal ECM receptors. It is likely that other proteins such as fibrinogen, colla- gen, and vitronectin also play roles which vary in im- portance from case to case depending on the extent to which plasma proteins are deposited on the wound bed.

Since there is no detectable TN among the epidermal cells of adult rats (Mackie et al., 1988) or adult humans (Lightner et al., 1989), it is somewhat surprising and interesting that normal unwounded urodele epidermis contains an abundance of TN. The function of this epi- dermal TN is unknown. Its adhesive properties suggest that it may simply be there as an intercellular cement. Possessing as it does six subunits, it has the necessary multiplicity of binding sites to join adjacent cells. The rapid loss of TN following wounding could explain why migrating epidermal cells are able to change their rel- ative positions in the sheet as a wound epithelium forms (Mahan and Donaldson, 1986; Repesh and Ober-

456 D.J. DONALDSON ET AL.

Figs. 4-5

TENASCIN AND SKIN WOUNDS 457

(Figs.4, 5) Fig. 4. Cryostat sections through skin and underlying tissues from a sample taken 24 hr post-wounding (type I1 wound). Panel A is a section stained with H & E. The lower two panels show the results of staining an adjacent section with a monoclonal antibody to TN. In panel A, the wound is to the right of the dotted line. The rectangles on the left and right show the approximate locations of the regions depicted in panels B and C, respectively. In panel B, note that all TN immunoreactivity has disappeared from the uninjured epidermis (ue) adjacent to the wound. In panel C, there is likewise no TN evident in the wound epithelium (we). Deep to the wound surface, occasional sections included immunolabeled tendon (t), which is pos- itive even in unwounded limbs. Infrequently, localized areas of a vac- uolated material (v) that appeared to contain TN were also seen deep to the wound epithelium. Otherwise, there was no labeling of the deep tissues. The arrows in both lower panels point to the dermo-epidermal junction. Magnification of A, 66 x ; B, 108 x ; C, 108 x .

Fig. 5. Cryostat sections through skin and underlying tissues from a sample taken 3 weeks after wounding (type I1 wound). Panel A is a section stained with H & E. The two lower panels show the results of staining an adjacent section with a monoclonal antibody to TN. In panel A, the left and right rectangles show the approximate locations of the regions depicted in panels B and C, respectively. Though the exact boundaries of the original wound are difficult to determine this long after wounding, the presence of glands and intact muscle under the epidermis enclosed by the left rectangle indicates that this repre- sents an area adjacent to the original wound. Panel B shows that the only TN staining here is in a length of tendon (t). Uninjured epider- mis (ue) is still negative. m = muscle. In the wound (panel C), TN is also still absent from the wound epithelium (we). The wound matrix (wm), however, is strongly TN-positive. Magnification of A, 66 x ; B, 108x; C, 1 0 8 ~ . Fig. 6. Cryostat sections through normal skin on an unwounded

limb and wound epithelium from a 24 hr type I wound, both of which were stained with a polyclonal antibody to TN. The normal skin on the left was taken from the contralateral limb of the animal providing the wound epithelium. Note that TN immunoreactivity is greatly re- duced in the wound epithelium (we) compared to normal epidermis he ) . g = glands, m = muscle. Magnification, 125 x .

100

80

60

40

20

0

COLL TN TN MY0 MY0 2001250 100 250 100 250

Conc. (pg/ml) Fig. 7. The epidermal response to TN as a migration substrate. Skin

explants were cultured for 16 hr in dishes coated with the indicated proteins. An ELISA assay of dishes coated with various concentra- tions of TN indicated that TN binding saturated a t 250 pgiml. The

extent of epidermal migration in each group was determined by planimetry. N = 5 for myoglobin a t 100 pgiml; N 2 6 for myoglobin a t 250 pgiml and for both TN groups. The lines on each bar indicate the S.E.

458 D.J. DONALDSON ET AL.

priller, 1980). Alternatively, it could be that uninjured epidermal cells normally interact with the anti-adhe- sive domain of TN, a situation which could promote structural stability of the epidermis by inhibiting po- tentially disruptive interactions with other adhesive proteins. In this case, the loss of TN following wound- ing would allow epidermal cells to interact with migra- tion-promoting proteins in the basement membrane and the wound. The loss of TN could be accomplished by the proteolytic events that are known to occur fol- lowing wounding. Interestingly, we have found that certain protease inhibitors block migration (unpub- lished data). Whether this is related in any way to TN retention following wounding remains to be estab- lished.

When pieces of newt skin were placed in TN-coated dishes, epidermal migration was quite poor, suggesting that even if a small amount of TN was present in the provisional wound matrix, it would not likely be among the proteins providing a suitable substrate for wound closure. This conclusion carries a slight caveat, name- ly, a reminder that by necessity heterologous TN was used to coat the dishes in these experiments. While this fact should be kept in mind, it should also be noted that heterologous fibronectin, fibrinogen, collagen, and vit- ronectin have all been shown to promote abundant mi- gration of newt epidermal cells in this in vitro system (Donaldson et al., 1987). Similarly, Probstmeier et al. (1990) reported that the intestinal epithelial cell line HT-29, derived from a human colon adenocarcinoma, adheres to substratum-immobilized mouse laminin, various non-human collagens (rat type I, chick type 11, bovine type 111, and mouse type IV) but not to mouse JlITN.

The results presented here also show that the loss of epidermal TN and its widespread expression in mesen- chymal tissues of the newt do not depend on the epige- netic mechanisms that regulate appendage regenera- tion. Simple wounding is sufficient to cause both events. The function of TN in healing mesenchymal tissues is as problematic as in normal epidermis. Since the cells in these two compartments are so different in other ways, they may also react differently to TN. Though it does not appear to mediate migration of epi- dermal cells, it could do so for mesenchymal cells. Al- ternatively, i t could modulate the interaction of these cells with other ECM proteins such as FN (Tan et al., 1987; Chiquet-Ehrismann et al., 19881, optimizing them as migration substrates rather than minimizing them as we have suggested for epidermis. Or i t may act as a mitogen (Chiquet-Ehrismann et al., 1986). What- ever its role in the epidermis or mesenchymal tissues, the dramatic shift in its expression following wounding suggests that it exerts a major influence in both places.

ACKNOWLEDGMENTS This study was supported by NIH grants AR27940,

BRSG RR05423 and DK04256.

LITERATURE CITED Arsanto, J.-P., M. Diano, Y. Thouveny, J.P. Thiery, and G. Levi 1990

Patterns of tenascin expression during tail regeneration of the amphibian urodele Pleurodeles waltl. Development, 109:177-188.

Bourdon, M.A., and E. Ruoslahti 1989 Tenascin mediates cell attach- ment through an RGD-dependent receptor. J. Cell Biol., 108: 1149-1155.

Bourdon, M.A., T.J. Matthews, S.V. Pizzo, and D.D. Bigner 1985 Im- munochemical and biochemical characterization of a glioma-as- sociated extracellular matrix glycoprotein. J. Cell Biochem., 28: 183-195.

Chiquet, M., and D.M. Fambrough 1984a Chick myotendinous anti- gen. I. A monoclonal antibody as a marker for tendon and muscle morphogenesis. J. Cell Biol., 98t1926-1936.

Chiquet, M., and D.M. Fambrough 1984b Chick myotendinous anti- gen. 11. A novel extracellular glycoprotein complex consisting of large disulfide-linked subunits. J. Cell Biol., 98t1937-1946.

Chiquet-Ehrismnnn, R., E.J. Mackie, C.A. Pearson, and T. Sakakura 1986 Tenascin: An extracellular matrix protein involved in tissue interactions during fetal development and oncogenesis. Cell, 47; 131-139.

Chiquet-Ehrismann, R., P. Kalla, C.A. Pearson, K. Beck, and M. Chi- quet 1988 Tenascin interferes with fibronectin action. Cell, 53: 383-390.

Clark, R.A.F., J.M. Lanigan, P. DellaPelle, E. Manseau, H.F. Dvorak, and R.B. Colvin 1982 Fibronectin and fibrin provide a provisional matrix for epidermal cell migration during wound reepithelial- ization. J. Invest. Dermatol., 79t264-269.

Donaldson, D.J., and J.T. Mahan 1983 Fibrinogen and fibronectin as substrates for epidermal cell migration during wound closure. J. Cell Sci., 62.117-127.

Donaldson, D.J., and J.T. Mahan 1984 Epidermal cell migration on laminin-coated substrates: Comparison with other excracellular matrix and non-matrix proteins. Cell Tissue Res., 235t221-224.

Donaldson, D.J., J.T. Mahan, and G.N. Smith, J r . 1987 Newt epider- mal cell migration in uitro and zn uiuo appears to involve Arg- Gly-Asp-Ser receptors. J. Cell Sci., 87t525-534.

Donaldson, D.J., J.T. Mahan, and G.N. Smith, J r . 1988 Newt epider- mal cell migration over collagen and fibronectin involves differ- ent mechanisms. J. Cell Sci., 90t325-333.

Erickson, H.P., and M.A. Bourdon 1989 Tenascin: An extracellular matrix protein prominent in specialized embryonic tissues and tumors. Annu. Rev. Cell Biol., 55'-92.

Erickson, H.P., and J.L. Iglesias 1984 A six-armed oligomer isolated from cell surface fibronectin preparations. Nature, 311 t267-269.

Friedlander, D.R., S. Hoffman, and G.M. Edelman 1988 Functional mapping of cytotactin: Proteolytic fragments active in cell-sub- strate adhesion. J. Cell Biol., 107t2329-2340.

Grumet, M., S. Hoffman, K.L. Crossin, and G. M. Edelman 1985 Cy- totactin, an extracellular matrix protein of neural and non- neural tissues that mediates glia-neuron interaction. Proc. Natl. Acad. Sci. U.S.A.,82:8075-8079.

Halfter, W., R. Chiquet-Ehrismann, and R.P. Tucker 1989 The effect of tenascin and embryonic basal lamina on the behavior and mor- phology of neural crest cells in uitro. Dev. Biol., 132t14-25.

Hoffman, S., K.L. Crossin, and G. Edelman 1988 Molecular forms, binding functions, and developmental expression patterns of cy- totactin and cytotactin-binding proteoglycan, an interactive pair of extracellular matrix molecules. J. Cell Biol., 106t519-532.

Kruse, J., G. Keilhauer, A. Faissner, R. Timpl, and M. Schachner 1985 The J1 glycoprotein-A novel nervous system cell adhesion molecule of the L2iHNK-1 family. Nature, 316t146-148.

Lightner, V.A., F. Gumkowski, D.B. Bigner, and H.P. Erickson 1989 Tenascinihexabrachion in human skin: Biochemical identifica- tion and localization by light and electron microscopy. J. Cell Biol., Z08t2483 -2493.

Mackie, E.J., R. Chiquet-Ehrismann, C.A. Pearson, Y. Inaguma, K. Taya, Y. Kawarada, and T. Sakakura 1987 Tenascin is a stromal marker for epithelial malignancy in the mammary gland. Proc. Natl. Acad. Sci. U.S.A., 84t4621-4625.

Mackie, E.J., W. Halfter, and D. Liverani 1988 Induction of tenascin in healing wounds. J. Cell Biol., 107t2757-2767.

Mahan, J.T., and D.J. Donaldson 1986 Events in the movement of newt epidermal cells across implanted substrates. J. Exp. Zool., 237:35-44.

Onda, H., D.J. Goldhamer, and R.A. Tassava 1990 An extracellular matrix molecule of newt and axolotl regenerating limb blastemas and embryonic limb buds: Immunological relationship of MT1 antigen with tenascin. Development, 108t657-668.

Probstmeier, R., R. Martini, and M. Schachner 1990 Expression of Jlitenascin in the crypt-villus unit of adult mouse small intes- tine: Implications for its role in epithelial cell shedding. Devel- opment, I09:3 13 -321.

Repesh, L.A., and J.C. Oberpriller 1980 Ultrastructural studies on migrating epidermal cells during the wound healing stage of re-

TENASCIN AND SKIN WOUNDS 459

generation in the adult newt, Notophthalrnus uiridescens. Am. J. Anat., 159:187-208.

Spring, J., K. Beck, and R. Chiquet-Ehrismann 1989 Two contrary functions of tenascin: Dissection of the active sites by recombi- nant tenascin fragments. Cell, 59,325-334.

Tan, S.S., K.L. Crossin, S. Hoffman, and G . Edelman 1987 Asymmet- ric expression in somites of cytotactin and its proteoglycan ligand is correlated with neural crest cell distribution. Proc. Natl. Aca. Sci. U.S.A., 84t7977-7981.