Adhesion Complex Formation After Small Keratectomy Wounds in the Cornea

E. Lee Stock,*t Michelle A. Kurpakus4 Barbara Sambol,*t and Jonathan C. R. JonesJ

The adhesion complex of the corneal epithelium consists of the hemidesmosome and its associated structures, such as anchoring filaments, lamina densa of the basement membrane, and anchoring fibrils. It contributes to the adhesion of the corneal epithelium to Bowman's layer. To understand the adhesion complex better, an electron microscopic and immunofluorescence analysis was done of the reformation of the adhesion complex in small (1 mm) keratectomy wounds in the guinea pig cornea. In these wounds, the epithelium, hemidesmosomes, basal lamina, anchoring fibrils, and anterior stroma were removed. The wound bed was epithelialized completely by 24 hr after wounding. Immunofluores- cence analyses involved the use of antibodies against plaque components of the hemidesmosome, an antibody against laminin, and an antibody against the collagen VII component of anchoring fibrils. At 18 hr after wounding, there was no morphologic evidence of hemidesmosomes at the epithelial-stromal interface. At 24 hr, hemidesmosomes were observed, with or without subjacent lamina densa. Further- more, plaque components were detected by immunofluorescence in those cells in contact with the wound bed. In contrast, no type VII collagen was detected. On day 7, collagen VII, laminin, and bullous pemphigoid autoantibody markers colocalized along the wound bed as determined by immunofluorescence. However, at the ultrastructural level, even though the lamina densa of the basal lamina was observed primarily where hemidesmosomes were present, it remained incomplete. In this study, the precise temporal sequence in which components are incorporated into the assembling adhesion complex was described during wound healing. Furthermore, the possibility that the hemidesmosomal plaque nucleates the formation of the underlying basal lamina was discussed. Invest Ophthalmol Vis Sci 33:304-313,1992

The corneal epithelium adheres to the underlying to be understood. For example, recent studies have stroma in part by the adhesion complex, consisting of revealed that two high molecular weight polypeptides a hemidesmosome, anchoring filaments that traverse (180 and 230 kD) recognized by autoantibodies in the the lamina lucida region of the basal lamina, and an- serum of patients with the blistering disease bullous choring fibrils that arise at the lamina densa of the pemphigoid (BP) are located in the plaque of the he- basal lamina and splay out into the stroma.12 On its midesmosome.3 Furthermore, both in the skin and cytoplasmic side, the hemidesmosome possesses a tri- cornea, it has been shown that the major component partite plaque.1 This plaque acts as the anchorage site of anchoring fibrils is type VII collagen.24 of bundles of intermediate filaments (IF). The bio- To study the assembly of hemidesmosomes, several chemical nature of the adhesion complex is beginning experimental systems have been used. For example, subepidermal blisters were created and hemidesmosome reformation was monitored as epithelial cells repopulated the denuded basal lamina.5 Others used From the *Cornea and External Eye Disease Laboratory, VA an in vitro system in which epithelial cell sheets were Lakeside Medical Center, and the Departments of fOphthalmology 6 and tCell Molecular and Structural Biology, Northwestern Univer- added back to denuded basal lamina. In this system, sity Medical School, Chicago, Illinois. hemidesmosome reformation is rapid, and it has been Supported in part by the Department of Veterans Affairs Re- proposed that hemidesmosome assembly is nucleated search Service (Washington, DC) grant 176-34-3874-03 (ELS) and by the preexisting anchoring fibrils associated with the National Institutes of Health (Bethesda, Maryland) grant GM- the basal lamina. 38470 to JCRJ. JCRJ is a Junior Faculty Research Awardee of the American Cancer Society (grant JFRA-232). MAK was a National In models where epithelial cells repopulate connec- Eye Institute (Bethesda, Maryland) postdoctoral fellow (grant EY- tive tissue in the absence of basal lamina, there are 06237). conflicting reports concerning the temporal sequence Submitted for publication: May 3, 1991; accepted September 17, of the reformation of the complete adhesion complex. 1991. Some authors found that hemidesmosomes, basal Reprint requests: E. Lee Stock, Cornea and External Eye Disease 7 Laboratory, Northwestern University Medical School, 303 East lamina, and anchoring fibrils appear concomitantly. Chicago Avenue, Chicago, IL 60611. Others report that the hemidesmosome plaque assem-

304

Downloaded from iovs.arvojournals.org on 09/28/2021 No. 2 ADHESION COMPLEX FORMATION IN THE CORNEA / Stock et ol 305

bles before the appearance of the lamina densa region We therefore wished to determine whether this is also of the basal lamina.8 the case in vivo. We analyzed the reformation of In previous studies, it has been shown that, in an in hemidesmosomes, anchoring fibrils, and basal lamina vitro model of wound healing, the plaque compo- in small (1 mm) keratectomy wounds in vivo in the nents of the hemidesmosome appear, in most in- guinea pig cornea by electron microscopy and immu- stances, before the appearance of collagen VII.910 Fur- nofluorescence analysis using BP autoantibodies as thermore, the lamina densa region of the basal lamina markers for the hemidesmosome plaque, a monoclo- appears to form subsequently and immediately subja- nal antibody against collagen VII, and an antibody cent to forming hemidesmosomal plaque structures.9 against laminin.

Fig. 1. (A) Light micrograph of keratectomy wound immediately after wounding. (B) Light micrograph of keratectomy wound 24 hr after wounding (XI400).

Downloaded from iovs.arvojournals.org on 09/28/2021 306 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / February 1992 Vol. 33

Materials and Methods Medical School (Chicago, IL). The anticollagen VII 9 Keratectomies monoclonal antibody was characterized previously. Antilaminin antibody was purchased from Telios All guinea pigs were anesthetized with xylazine HC1 (San Diego, CA). (8 mg/kg) and ketamine HC1 (120 mg/kg), in accor- dance with the ARVO Resolution on the Use of Ani- Light Microscopic Analysis mals in Research. Topical tetracaine 0,5% (four Cryostat sections were cut at 6 nm and placed on drops) was used to supplement the anesthetic. A 1- polylysine-coated (Sigma, St. Louis, MO) slides. Dou- mm Elliot trephine (Storz, St. Louis, MO) was used ble-label immunofluorescence was done as detailed for the keratotomy, and the keratectomy was com- previously.9 In brief, sections on slides were fixedfo r 5 pleted with a sharp blade. Chloramphenicol ophthal- min in -20°C acetone and then air dried. A mixture mic ointment (0.5%) was placed on the eye, and the of primary antibodies diluted in phosphate-buffered animals were allowed to recover. The animals were saline (PBS) was overlaid on the sections. The slides killed by an overdose of sodium pentobarbital at 18, were incubated 1 hr at 37°C, washed in PBS, and 24, and 48 hr or 7 days, and the corneas were re- overlaid with a mixture of appropriate secondary fluo- moved. The corneas were either frozen solid in liquid rochrome-conjugated antibodies. After washing in nitrogen, then embedded in O.C.T. compound PBS, cover slips were placed over sections which were (Miles, Inc., Elkhart, IN) forimmunofluorescence mi- viewed on a Leitz Diaplan microscope (Ernst Leitz, croscopic analysis, or fixedi n 1.0% glutaraldehyde for Wetzlar GMBH, Germany) with epifluorescence. electron microscopy. Electron Microscopy Antibodies After fixation in 1.0% glutaraldehyde for at least 2 Serum samples from patients with BP were pro- hr, the tissue was washed six times in PBS, placed in vided by Nancy Furey, MD, Northwestern University 1% osmium tetroxide in PBS for 90 min at room tern-

Fig. 2. Transmission electron micrograph of keratectomy wound at 18 hr. Note that there are no hemidesmosomes along regions of epithelial-stromal interaction or an obvious basal lamina (arrows). Intermediate filaments (IF) are present but are not associated with the basal surface of the cell (original magnification X60.000).

Downloaded from iovs.arvojournals.org on 09/28/2021 No. 2 ADHESION COMPLEX FORMATION IN THE CORNEA / Srock er ol 307

perature, and rinsed in distilled water three times for 1 Results minute each. The tissue was dehydrated in graded Keratectomy wounds of 1 mm resulting in the re- ethanols, and two 15-min changes of propylene oxide moval of both epithelium and basal lamina were were completed. It then was placed in 1:1 propylene made in the cornea. At various times after wounding, oxide-Epon/Araldite resin mix (Electron Microscopy \-nm sections of Epon-embedded corneas were pro- Sciences, Fort Washington, PA) for 2 days at room cessed for light microscopic observation. Immediately temperature for infiltration. On day 2, the cap was after wounding, the normal corneal epithelium ends loosened to allow propylene oxide to evaporate from abruptly on either side of the wound site (Fig. 1A). the mixture. Final embedding was accomplished in However, within 24 hr, epithelial cells have repopu- 100% Epon/Araldite resin, and the tissue was placed lated the wound bed completely, and the cells already in a 60°C dry oven for 48 hr. are stratified (Fig. IB). There is some evidence of in- Sections were cut at 1 ^tm on a Reichert-Jung Ultra- flammation in the wound site (Fig. IB). No obvious cut E microtome (Reichert, Buffalo, NY) and stained differences were seen between wounds at 24 hr and 7 with toluidine blue. The wound was localized in the days at the light microscopic level of resolution (re- 1-Mm sections and the block trimmed for thin-section sults not shown). localization. Ultrathin sections were cut at 80 nm, placed on 300-mesh copper grids, stained with uranyl Wound Healing at 18 Hr acetate for 20 min and lead citrate for 1 min, rinsed in At 18 hr after wounding, where epithelial cells had filtered distilled water, and viewed on a JEOL 100CX moved over the wound bed, no obvious hemidesmo- (Peabody, MA) electron microscope at 60 kV. somes or basal lamina were observed along the area of

Fig. 3. lmmunofluorescence analyses of cryostat sections of wounded corneas 18 hr postkeratectomy. (A) and (B) are double-labeled with collagen VII antibodies and BP autoantibodies, respectively. (C) and (D) are double-labeled with laminin antibodies and BP autoantibodies, respectively. No staining is observed along the wound bed with collagen type VII antibody preparations (A), or laminin (C), and very little staining is seen with BP (B, D, closed arrows). However, these antibodies generate linear staining along the epithelial-stromal (E = epithelial, S = stroma) interface of the unwounded ocular surface (open arrows) (original magnification X1200).

Downloaded from iovs.arvojournals.org on 09/28/2021 308 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / February 1992 Vol. 30

epithelial-stromal interface as determined by electron 24 hr postkeratectomy (Fig. 4). The IF were asso- microscopy (Fig. 2). The IF appear disorganized and ciated with these electron-dense plaques (Fig. 4A). In did not associate obviously with the basal cell surface some instances, collagen fibers in the stroma were (Fig. 2). Those collagen fibersi n the stroma that abut- aligned perpendicular to the hemidesmosomes in the ted the basal surface of the epithelial cell were aligned epithelial cells in the absence of any obvious basal perpendicular to the epithelial-stromal interface (Fig. lamina (Fig. 4A). However, we also have observed 2). Double-label immunofluorescence analyses of lamina densa immediately subjacent to hemidesmo- cryostat sections of wounded corneas show that, at 18 somes in other areas of the same wound (Fig. 4B). In hr postkeratectomy, little staining was observed along the latter, the hemidesmosomes were similar in mor- the wound bed with BP, laminin, and collagen type phology to those seen in normal areas; they possessed VII antibody preparations, although these antibodies a tripartite plaque to which organized bundles of IF caused linear staining along the epithelial-stromal in- were attached (Fig. 4B). terface of the unwounded ocular surface (Fig, 3). Double-label immunofluorescence observations of corneas at 24 hr after wounding revealed that BP anti- Wound Healing at 24 Hr gens occurred along the wound bed; no staining was Hemidesmosomal plaque structures were present detected in this region using the collagen VII or la- along the epithelial-stromal interface of the wound at minin antibody preparations (Fig. 5).

Fig. 4. Transmission electron micrograph of the wound bed at the epithelial-stromal interface 24 hr after wounding. (A) Hemidesmosomes are present with their sub-basaJ dense plate, but there is no obvious basal lamina (arrows) IF, intermediate filaments (original magnification X63,OOO).

Downloaded from iovs.arvojournals.org on 09/28/2021 No. 2 ADHESION COMPLEX FORMATION IN THE CORNEA / Stock et ol 309

Fig. 4. (B). The lamina densa of the basal lamina is immediately subjacent to each hemidesmosome (arrows). Note that there are no areas of lamina densa without overlying hemidesmosomes. The hemidesmosomes have a tripartite plaque structure to which intermediate filaments (IF) attach (original magnification X30,000).

Wound Healing at 48 Hr Double-label immunofluorescence of the wound site on day 7 revealed that BP autoantibodies and col- Hemidesmosomes with an underlying lamina lagen VII antibodies now caused similar punctate densa were observed along the region of epithelial- staining patterns along the wound bed (Figs. 7A-B). stromal interface at 48 hr after wounding (results not By contrast, although laminin colocalized with BP shown); this was similar to that observed at 24 hr (Fig. antigens along the epithelial-stromal interface of the 4B). Double-label immunofluorescence analyses of wound bed, the laminin antibodies also caused a dif- wounded corneas revealed that, along the epithelial- fuse stain extending into the stroma immediately un- stromal interface of the wound bed, both BP and la- derlying the wound bed (Figs. 7C-D). minin antibodies colocalized. No staining was observed using the collagen type VII antibody prepara- tion (Fig. 6). Discussion We did a detailed immunofluorescence and elec- Wound Healing at 7 Days tron microscopic analysis of assembly of the adhesion The epithelial-stromal interface 7 days after kera- complex in small (1 mm) keratectomy wounds in the tectomy did not differ obviously from that observed guinea pig cornea. In such small wounds, reepithelia- in corneas 24 hr after wounding (Fig. 4B). There were lization occurs much more rapidly than in the large (7 many areas where the lamina densa region of the ba- mm) keratectomy wounds in rabbit cornea used by sal lamina was discontinuous and only occurred im- other workers in comparable studies.7 In the larger mediately underlying hemidesmosomes (results not wounds, reepithelialization is not complete at 48 hr shown). even though hemidesmosome assembly has begun.

Downloaded from iovs.arvojournals.org on 09/28/2021 310 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / February 1992 Vol. 33

Fig. 5. Immunofluorescence analysis of the wound bed 24 hr after wounding. (A) and (B) are double-labeled with collagen VII antibodies and BP autoantibodies, respectively. (C) and (D) are double-labeled with laminin antibodies and BP autoantibodies, respectively. In (A), collagen VI1 antibody stains the interface between the epithelium (E) and stroma (S) of the unwounded cornea (arrow), but there is no staining in the wound bed. In (B), BP antibodies stain the interface between the unwounded epithelium (E) and stroma (S) (open arrow), and also generate intermittent staining along the wound bed (closed arrow). In (C), laminin stains the interface between the epithelium and stroma of the unwounded cornea, but there is no staining in the wound bed. In (D), there is staining with BP antibodies in the unwounded cornea, as well as the wound bed (original magnification X1200).

Analyses of adhesion complex formation in such wound healing in the cow where the appearance of wounds is complicated by the possibility that epithe- laminin preceded that of hemidesmosomal plaque lial cells are still migratory. In our study, we were able components.9 It is possible that in vitro models may to correlate precisely the appearance of adhesion not simulate entirely the in vivo situation. For exam- complex components immunocytochemically with ple, differences between in vivo and in vitro models the appearance of hemidesmosomes at the ultrastruc- may reflect the inflammatory response seen in vivo. tural level in cells that were not migratory because the The latter may result in degradation of certain extra- wound site was covered completely. cellular matrix proteins by specific metalloprotein- Our results showed that, during the reepithelializa- ases.11 Alternatively, there may be differences be- tion of i-mm keratectomy wounds in vivo, hemides- tween wound healing in guinea pig compared with mosomal plaque components appeared at the epithe- cow ocular tissue (which are distinct morphologi- lial-stromal interface of the wound within 24 hr. At cally). approximately the same time, hemidesmosomal Our immunocytochemical results show that hemi- plaques could be observed ultrastructurally simulta- desmosomal plaque components appear before colla- neously with the appearance of BP antigen. Laminin gen VII during adhesion complex formation. These occurs along the epithelial-stromal interface of the results differ from those presented for wound healing wound bed at 24-48 hr. This contrasts with the results in the rabbit cornea where hemidesmosomal plaque obtained recently using an in vitro model of corneal components are claimed to appear simultaneously

Downloaded from iovs.arvojournals.org on 09/28/2021 No. 2 ADHESION COMPLEX FORMATION IN THE CORNEA / Srock er ol 311

Fig. 6. Tmmunofluorescence analysis of the wound bed 48 hr after wounding. (A) and (B) are double-labeled with collagen VII antibodies and BP autoantibodies, respectively. (C) and (D) are double-labeled with laminin antibodies and BP autoantibodies, respectively. In (A), collagen VII antibody stains at the interface between the epithelium (E) and stroma (S) in the normal cornea (open arrow), but there is no staining in the wound. In (B), the unwounded area stains with BP autoantibodies (open arrow), and these autoantibodies generate a punctate stain along the wound bed (closed arrow). In (C), laminin antibodies stain the normal part of the cornea (open arrow), and along the wound bed in a similar pattern to BP autoantibodies (compare C and D) (original magnification X1200).

with collagen VII.12 However, these authors did not be nucleated by preexisting anchoring fibrils.6 How- study the earliest stages of hemidesmosome assembly ever, if the basal lamina (and anchoring fibrils) is ab- (ie, 18-24 hr after wounding) as we did but rather sent, as is the case in a keratectomy wound, hemides- presented data for adhesion complex formation at 48 mosomes form first with later anchoring fibril forma- hr, a time when newly formed hemidesmosomes and tion. Recent evidence suggests that collagen VII is hot underlying basal lamina already can be observed ultra- required for the development of hemidesmosomes. It structurally. was shown that sheep with recessive dystrophic epi- Others suggested that anchoring fibrils appear to dermolysis bullosa do not possess collagen VII, yet nucleate the assembly of the hemidesmosome in a still have morphologically intact hemidesmosomes.13 recombination model where corneal epithelium is The possibility of hemidesmosomes being laid reassociated with denuded stroma with an intact basal down first, followed by basal lamina, is attractive be- lamina.6 However, in our study, the appearance of cause our ultrastructural studies revealed that the lam- collagen type VII, an anchoring fibrilcomponent , fol- ina densa region of the basal lamina appeared immedi- lowed the appearance of both plaque components of ately underlying hemidesmosomal plaques during the hemidesmosome and laminin. Therefore, we wound healing. We suggest a role for hemidesmo- found no evidence that anchoring fibrils nucleate somes in organizing or nucleating formation of the hemidesmosome assembly. We are tempted to ex- basal lamina and its associated structures, such as an- plain these results by hypothesizing that, if the basal choring fibrils.Thi s hypothesis is supported by several lamina is present, hemidesmosome reformation can reports in the literature. For example, it was found

Downloaded from iovs.arvojournals.org on 09/28/2021 312 INVESTIGATIVE OPHTHALMOLOGY b VISUAL SCIENCE / February 1992 Vol. 33

Fig. 7. Immunofluorescence analysis of the wound at 7 days. (A) and (B) are double-labeled with collagen VII antibodies and BP autoantibodies, respectively. (C) and (D) are double-labeled with laminin antibodies and BP autoantibodies, respectively. In (A), collagen VII antibodies generate intermittent staining along the interface between the stroma (S) and the epithelium (E) in the region of the wound (closed arrows). In the same section shown in (B), BP autoantibodies show intense staining all along the same interface (solid arrows). (C) Laminin contrasts with the other antibodies since it is deposited diffusely in the stroma underneath the wound bed (solid arrows). Compare this with the linear deposition of BP antigens shown in the same section (D, solid arrows) (original magnification X1200).

that basal lamina formed under the hemidesmosomes 2. Gipson IK, Spurr-Michaud SJ, and Tisdale AS: Anchoring fi- when human epidermis was recombined with human brils form a complex network in human and rabbit cornea. dermis devoid of its basal lamina.8 This sequence does Invest Ophthalmol Vis Sci 28:212, 1987. 3. KJatte DH, Kurpakus MA, Grelling KA, and Jones JCR: Im- not occur during development; therefore, there are munochemical characterization of three components of the he- differences between development and healing. midesmosome and their expression in cultured epithelial cells. In summary, we described the sequence of incorpo- J Cell Biol 109:3377, 1990. ration of three components into the assembling adhe- 4. Sakai LY, Keene DR, Morris NP( and Burgeson RE: Type VII sion complex. Because recent studies have shown that collagen is a major structural component of anchoring fibrils. J some extracellular matrix receptors belonging to the Cell Biol 103:1577, 1986. 1415 5. Krawczyk WS and Wilgram GF: Hemidesmosome and desmo- integrin family occur in the hemidesmosome, it some morphogenesis during epidermal wound healing. J UI- will be interesting to assess the role these receptors trastruct Mol Struct Res 45:93, 1073. play in hemidesmosome assembly. 6. Gipson IK, Grill SM, Spun SJ, and Brennan SJ: Hemidesmo- some reformation in vitro. J Cell Biol 97:489, 1983. Key words: cornea, wound heating, hemidesmosome, adhe- 7. Gipson IK, Spurr-Michaud S, Tisdale A, and Keough M: Reas- sion complex sembly of the anchoring structures of the corneal epitheliuln during wound repair in the rabbit. Invest Ophthalmol Vis Sci References 30:425, 1989. 8. Briggaman RA, Dalldorf FG, and Wheeler CE Jr: Formation I. Staehelin LA: Structure and function of intercellular junctions. and origin of basal lamina and anchoring fibrils in adult human Int Rev Cytol 39:191, 1974. skin. J Cell Biol 71:384, 1971.

Downloaded from iovs.arvojournals.org on 09/28/2021 No. 2 ADHESION COMPLEX FORMATION IN THE CORNEA / Stock er ol 313

9. Kurpakus MA, Stock EL, and Jones JCR: Analysis of wound somes and anchoring fibrilcollage n appear synchronously dur- healing in an in vitro model: Early appearance of laminin and a ing development and wound healing. Dev Biol 126:253, 1988. 125 X 103 Mr polypeptide during adhesion complex forma- 13. Bruckner-Tuderman L, Guscetti F, and Ehrensperger F: Ani- tion. J Cell Sci 96:651, 1990. mal model for dermolytic mechanobullous disease: Sheep with 10. Kurpakus MA and Jones JCR: A novel hemidesmosomal recessive dystrophic epidermolysis bullosa lack collagen VII. J plaque component: Tissue distribution and incorporation into Invest Dermatol 96:452, 1991. assembling hemidesmosomes in an in vitro model. Exp Cell 14. Stepp MA, Spurr-Michaud S, Tisdale A, Elwell J, and Gipson Res 194:139, 1991. IK: oifjlii integrin heterodimer is a component of hemidesmosomes. Proc Natl Acad Sci U S A 87:8970, 1990. 11. Matrisian LM: Metalloproteinases and their inhibitors in ma- 15. Jones JCR, Kurpakus MA, Cooper HM, and Quaranta V: A trix remodeling. Trends in Genetics 6:121, 1990. function for the integrin a6(t4 in the hemidesmosome. Cell Reg- 12. Gipson IK, Spurr-Michaud SJ, and Tisdale AS: Hemidesmo- ulation 2:427, 1991.

Downloaded from iovs.arvojournals.org on 09/28/2021