radial keratotomy. ii. role of the myofibroblast in...

Investigative Ophthalmology & Visual Science, Vol. 33, No. 12, November 1992Copyright © Association for Research in Vision and Ophthalmology

Radial Keratofomy

//. Role of rhe Myofibroblosr in Corneol Wound Conrrocrion

Randa M. R. Gorono,* W. Matthew Petroll,* Wen-Tien Cheat Ira M. Herman,:}: Patricia Barry,*Peter Andrews,-]- H. Dwight Cavanagh,t and James V. Jesterf

The cellular mechanism of corneal wound contraction after radial keratotomy (RK) was studied in afeline eye model. A total of 10 cat eyes were evaluated at various times from 0-30 days after surgery.Changes in the distribution of intracellular filamentous actin, nonmuscle myosin, a-actinin, surfacemembrane a5/?, integrin, and extracellular fibronectin were studied using immunofluorescence andlaser confocal and electron microscopy. From day 3-7, staining for fibronectin increased along thewound margin. By day 7, keratocytes adjacent to the wound margin showed increased f-actin stainingwith intense staining for fibronectin compared with normal keratocytes. Myosin and a5/?, integrinexpression was very weak at this time; a-actinin was not found. By day 14, fibroblasts within the woundformed f-actin microfilament bundles (stress fibers) which colocalized with fibronectin. Wound-healingfibroblasts also stained positively for a5j3l integrin, myosin, and a-actinin (the latter two were colocal-ized). The presence of myosin and a-actinin in the wound fibroblasts and the re-organization of f-actininto stress fibers by day 14 correlated with the development of wound contraction. A comparison of thecellular distribution of actin, myosin, and a-actinin with as01 integrin 14 days after injury suggestedthat integrin was localized along stress fiber bundles during wound contraction. The data from thisstudy suggest that modulation of wound gape during healing of RK wounds may involve transformationof the corneal keratocyte to a myofibroblast-like cell and the subsequent formation of intracellularstress fibers composed of f-actin, nonmuscle myosin, and a-actinin. Based on the colocalization offibronectin filaments and f-actin filaments and the unique distribution of a^ , integrin, these findingssupport the hypothesis that the tension within the wound is generated by the formation of intracellularstress fibers and the interactions between stress fibers and the extracellular matrix, mediated byspecific membrane receptor molecules. Invest Ophthalmol Vis Sci 33:3271-3282,1992

Recent studies of corneal wound healing after ra-dial keratotomy (RK) have shown that cornealwounds undergo a biphasic change in wound gape,including a precontractile and contractile phase.1 Thecellular mechanism of wound contraction has beenextensively studied in the skin. In 1971, work byMajno et al2 and Gabbiani et al3 identified a modifiedfibroblast that appears transiently in granulation tis-sue and has ultrastructural features intermediate be-tween those of the fibroblast and smooth muscle cell.

From the *Center For Sight and the tDepartment of Anatomyand Cell Biology, Georgetown University Medical Center, Washing-ton, DC, and the ^Department of Anatomy and Cell Biology, TuftsUniversity School of Medicine, Boston, Massachusetts.

Supported by in part by a grant from the National Institutes ofHealth (Bethesda, MD) EYO-7348, the Eleanor Naylor Dana Chari-table Trust, New York, New York, and an unrestricted grant fromResearch to Prevent Blindness, Inc. (New York, NY).

Submitted for publication: December 6, 1991; accepted May 21,1992.

Reprint requests: James V. Jester, Department of Ophthalmol-ogy, University of Texas Southwestern Medical Center, 5323 HarryHines Boulevard, Dallas, TX 75325.

Since these first reports, the role of this cell in woundcontraction has been demonstrated in various organsystems, including skin,2"4 liver,56 kidney,78 andlung.9 Furthermore, studies suggest that the origin ofthe myofibroblast may have diverse sources, such assmooth muscle cells, fibroblasts, pericytes, and macro-phages.910

Interest in corneal wound contraction has devel-oped after the introduction of keratorefractive sur-gery, particularly RK, where regression of the centralcorneal-flattening effects of this surgery has been at-tributed to a putative contractile mechanism of cor-neal wound healing." Contractility of cornealwounds was first demonstrated by Luttrull et al12 whoshowed that corneal fibrotic tissue contracted and re-laxed in response to selected pharmacologic agents.Further ultrastructural and biochemical studies re-vealed that corneal keratocytes undergo markedchanges in the intracellular distribution of filamen-tous actin with the formation of intracellular stressfibers characteristic of myofibroblast-like cells.13

Although myofibroblasts have been demonstrated

3271

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933386/ on 05/07/2018

3272 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / November 1992 Vol. 33

in full-thickness corneal wounds in the rabbit, thepresence of contractile cells in partial-thickness RK-produced wounds remains unconfirmed. Earlier de-tailed electron microscopic studies have failed to dem-onstrate the presence of activated fibroblasts with ul-trastructural characteristics of myofibroblasts."'14'15

This report documents the transformation of the cor-neal keratocytes to a myofibroblast-like cell after non-penetrating RK-produced wounds in the cat cornea.The relationship between stress fiber formation in themyofibroblast and the temporal expression of themembrane receptor, a5jft, integrin, was also deter-mined. These data indicate that the initiation ofwound contraction after RK. in the cat is associatedwith the ingrowth of activated fibroblasts having char-acteristics consistent with myofibroblasts. Further-more, the development of intracellular stress fiberswithin these cells appears to be organized by interac-tions between the extracellular matrix fibronectin, thesurface membrane fibronectin receptor, a5/3, integrin,and intracellular actin and actin binding proteins,myosin and «-actinin.

Materials and Methods

A total of 10 cats (10 eyes) were used in these stud-ies. All experiments with these animals were carriedout according to the ARVO Resolution on the Use ofAnimals in Research. The animals were anesthetizedwith acepromazine (0.5 mg/kg), ketamine (15 mg/kg), lidocaine (0.5 mg/kg), and atropine (0.025 mg/kg) before surgery. Each cat eye received four radialnonpenetrating corneal incisions using a diamondknife set at 85% of corneal thickness as measuredby ultrasonic pachymetry (Corneo-gage III; Sona-gauge, Beachwood, OH). The animals were killed atvarious times from 0-30 days after surgery by the in-jection of concentrated pentobarbital. The corneaswere: (1) perfused in situ for light and electron micros-copy or (2) removed fresh for immunofluorescencemicroscopy.

Light Microscopy

After the animals were killed, their corneas werefixed by anterior chamber perfusion with 2.5% glutar-aldehyde in 0.1 mol/1 phosphate buffer, pH 7.3, for 10min. The corneas were then removed and fixed over-night. Tissue blocks containing radial incisions werethen dehydrated and embedded in JB-4 embeddingmedia (Polysciences Inc., Warrington, PA). Blockswere sectioned on the Reichert-Jung 2050 microtome(Leica, Deerfield, IL). Four-micron thick sectionswere then stained with hematoxylin and eosin.

Transmission Electron Microscopy

Tissue blocks, 3X3 mm, were fixed overnight andthen washed three times in 0.1 mol/1 phosphatebuffer, pH 7.2. Tissue was then postfixed with 1% os-mium tetroxide in 0.1 mol/1 phosphate buffer for 1 hr,dehydrated in graded ethanols, embedded in spur(Polysciences), and polymerized overnight in a 60°Coven. Ultramicrotome sections were cut and thenstained with uranyl acetate and lead acetate. Sectionswere then examined on JEOL (Peabody, MA) 1200EX transmission electron microscope.

Fluorescence Microscopy

Fresh corneal tissue was immediately embedded inOCT Compound (Lab-Tek, Naperville, IL), frozen inliquid nitrogen, and stored at —80°C until sectioned.Cryostat sections were air dried, extracted in frozenacetone (-20°C), and rehydrated in phosphate-buff-ered saline, pH 7. Sections were reacted with: (1) fluo-rescein isothiocyanate (FITC)-rhodamine labeledphallacidin (Molecular Probes, Eugene, OR); (2)FITC-labeled goat anti-human fibronectin (Cappel-Organon Teknika, Durham, NC); (3) rabbit anti-hu-man nonmuscle myosin;16 (4) monoclonal antialphaactinin (ICN, Lisle, IL); or (5) rabbit anti-human a5/5,integrin.1718 Fluorescence-labeled secondary antibod-ies included: (1) FITC-conjugated immunoglobulinG (Ig) fraction goat anti-rabbit IgG (Cappel-OrganonTeknika) and (2) FITC-conjugated IgG fraction goatanti-mouse IgG (Cappel-Organon Teknika). Frozensections were viewed on two different microscopes:(1) the Olympus (Lake Success, NY) BHS/BHT photo-microscope equipped with epifluorescence, and (2) theBio-Rad (Cambridge, MA) MRC 600 laser confocalmicroscope and imaging system. Images obtainedfrom the Bio-Rad confocal microscope were trans-ferred to a Gould/Vicom (Fremont, CA) IP9527 imageprocessor and Matrix (Orangeburg, NY) film recorder.

Results

Light and Electron Microscopy

The histopathologic findings of corneal wound heal-ing in the cat were similar to those already describedin the primate." From 0-3 days after initial injury,there was migration of epithelium into the wound toform an epithelial plug (Fig. 1 A) which persisted untilday 10. During this precontractile stage, previous invivo confocal microscopic studies showed an increasein the wound gape of RK-produced incisions in thecat.1 Initiation of the contractile stage of in vivo cor-neal wound healing on day 14 was characterized bythe gradual replacement of the epithelial plug with


No. 12 CORNEAL MYOFIDRO3LAST / Gorona er al 3273

Fig. 1. Light microscopy of cat corneal wounds at 3 (a), 14 (b), and 30 (c) days and TEM at 14 days (d) after radial keratotomy surgery. Notethat the corneal wound is initially covered by corneal epithelium forming an epithelial plug (a). Epithelium is later replaced by fibrotic tissue by14 days(b, arrow). During this precontractile phase of corneal wound healing the wound gape appears to increase {a, b, arrow heads). From day14 to day 30, corneal wounds show a decrease in wound gape, suggesting corneal wound contraction (b, c, arrowheads) with a concomitantdecrease in the number of fibroblasts. Transmission electron microscopy of 14-day corneal wound-healing fibroblasts revealed prominentstress fiber bundles (d, SF) throughout the cell. Stress fibers appeared to insert at the plasma membrane adjacent to extracellular collagen fibers(arrow) (a, XI25; b, XI60; c, X29O; d, XI 5,000).


3274 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / November 1992 Vol. 33

fibrotic tissue (Fig. 1B, arrow). A comparison of thewound gape observed at days 3-14 (arrow heads) wasconsistent with the earlier in vivo observation of in-creasing wound gape during the precontractile phaseof wound healing. During the contractile phase (days14-30), there was a progressive decrease in woundgape (Fig. 1C, arrow heads) with a concomitant de-crease in the cellularity of fibrotic tissue. Transmis-sion electron microcopy of RK-produced wounds cuten face revealed that wound-healing fibroblasts con-tained prominent microfilament bundles with elec-tron-dense plaques and stress fibers (Fig. ID, SF)which appeared to terminate adjacent to extracellularcollagen fibers (Fig. ID, arrow). Sectioning of inci-sional wounds in en face orientation provided the bestdemonstration of intracellular stress fibers; this mayin part explain why earlier studies have not identifiedthese structures.11'14'15

Immunofluorescence Microscopy

Fibronectin and f-actin: Immediately after injury,fibronectin appeared to codistribute with fibrinogenalong the wound margins (data not shown). However,by 3 days after injury, there was a marked increase inantifibronectin staining which was contiguous withthe wound margin and appeared to extend along thelamellar planes into the undamaged stroma (Fig. 2A,arrow). At the same time, FITC-labeled phallacidinstaining at 3 days showed prominent cortical stainingof the corneal epithelium with very weak staining ofthe adjacent keratocytes (Fig. 2D). By day 14, the epi-thelial plug had been replaced by developing fibrotictissue which were stained intensely by antifibronectinantibodies (Fig. 2B, arrow). Wound-healing fibro-blasts contained within the area of fibrosis alsostained intensely with FITC-labeled phallacidin (Fig.2E, arrow), confirming the presence of stress fibers asidentified by transmission electron microscopy. Boththe fibronectin and f-actin distribution appeared to bedisorganized at this stage of wound healing. Kerato-cytes adjacent and contiguous with the wound alsostained intensely with FITC-labeled phallacidin; how-ever, keratocytes farther removed from the woundcontinued to show only weak staining. During thecontractile phase of corneal wound healing, from days14-30, progressive organization of fibronectin and f-actin distribution into a lamellar orientation similarto the stromal lamellae was observed (Figs. 2C, 2F).The colocalization of f-actin and fibronectin revealedin the 14- and 30-day samples was confirmed by dou-ble-immunofluorescence labeling of 14-day woundssectioned en face and stained with rhodamine phalla-cidin and FITC-labeled anti-fibronectin (Fig. 3). Asection taken at the epithelial-wound interface

showed cortical rhodamine phallacidin staining of theepithelium and stress fiber staining in the adjacentwound fibroblasts (arrow). The same section stainedwith FITC-labeled antifibronectin revealed that fibro-nectin was organized into filaments that ran parallelto the f-actin stress fibers (Fig. 3B, arrow) but not tothe epithelial cortical actin network. Sections deeperwithin the wound showed a similar colocalization off-actin stress fibers (Fig. 3C, arrow) with fibronectinfilaments (Fig. 3D, arrow).

Myosin and a-actinin: Nonmuscle myosin and a-actinin are important actin-binding proteins that arelocalized along stress fiber f-actin bundles.19 In thenormal corneal keratocyte, which does not containstress fibers, nonmuscle myosin is localized to the cellcortex, whereas a-actinin does not appear to be ex-pressed.20 By 3 days after RK, antinonmuscle myosinappeared to stain more intensely at the margins of thewound (Fig. 4A, arrow) and was localized predomi-nantly adjacent to the base of the migrating cornealepithelial cells and the adjacent keratocytes (Fig. 4B).By day 14 (Fig. 4C), anti-nonmuscle myosin showedincreased staining of fibroblasts within the woundthat appeared to be distributed around the cortex ofthe cells (Fig. 4D, arrow). Staining around the fibro-blasts was not completely uniform, however; someregions appeared to exhibit brighter linear fluores-cence compared with other areas within the same cell.The contractile phase of wound healing from days14-30 showed an increasing organization of the non-muscle myosin distribution similar to that seen forf-actin. By day 30, nonmuscle myosin was organizedinto parallel lines or filaments which extended consid-erably beyond the wound margins (Figs. 4E-F,arrows). The distribution of a-actinin appeared simi-lar to that of nonmuscle myosin except that duringthe early precontractile phase of wound healing, kera-tocytes adjacent to the wound showed only weakstaining by antibodies to a-actinin (Figs. 5A-B,arrows). By day 14, fibroblasts within the woundshowed intense staining along the margins of the cellin a disorganized pattern similar to that observed formyosin and f-actin (Figs. 5B-C).

Anti-a-actinin staining became more organizedduring the contractile phase, showing a linear stainingpattern albeit less intense than that seen at day 14(Figs. 5E-F). The colocalization of a-actinin and non-muscle myosin was confirmed using double iramuno-labeling of 14-day wounds stained by rhodamine-con-jugated goat anti-mouse IgG to localize a-actinin (Fig.6A) and FITC-conjugated goat anti-rabbit IgG to lo-calize nonmuscle myosin (Fig. 6b). Wound-healingfibroblasts below the overlying epithelium showed al-most an identical nonmuscle myosin and a-actinindistribution, whereas in the corneal epithelium, the


No. 12 CORNEAL MYOFIDRODLAST / Gorono er ol 3275

Fig. 2. Cat radial keratotomy at 3 (a, d), 14 (b, e), and 30 (c, f) days after surgery stained with FITC-conjugated goat anti-human fibronectin(a-c) and FITC-phallacidin (d-f). Three days after surgery, anti-fibronectin antibodies (a) stain the stromal edge of the wound with stainingextending out along the stromal lamellar planes (arrow). At the same time, FITC-phallacidin (d) stains the cortex of the epithelial cells fillingthe wound, but only weakly stains adjacent keratocytes. By day 14, developing fibrotic tissue shows an intense, disorganized anti-fibronectin(b) and phallacidin (e) staining pattern (arrows). Additionally, keratocytes adjacent to the wound show increased staining, having a distributionsimilar to but not as extensive as that of anti-fibronectin. At day 30, anti-fibronectin (c) and phallacidin (f) staining appear organized intoparallel bundles (a-f, X200).

basal plasma membrane of the epithelial cells wasstrongly stained by anti-a-actinin and diffuselystained by antinonmuscle myosin (curved arrows).

Integrin and stress fiber formation: The high-affin-ity membrane receptor for fibronectin, the a50{ inte-grin,1718 was localized using rabbit polyclonal anti-human a5(3l integrin antibodies which react with the

a5 and /?, subunits. The normal cornea did not appearto stain with these antibodies; however, 3 days afterinjury, anti-a5j3t integrin staining was detected alongthe basal plasma membrane of the epithelial plug andalong cellular extensions from keratocytes adjacent tothe wound margin (Figs. 7A-B, arrows). The localiza-tion of a5/?| integrin taken with the localization of



Fig. 3. Double immunofluorescence labeling of a 14-day wound sectioned en face and stained with rhodamine phalloidin (a, c) and FITCanti-fibronectin (b, d). Sections were taken at the junction between epithelium and stromal interface (a, b) and deeper within the stromal

fibroblasts. Cortical f-actin in the basal corneal epithelial cells did not colocalize with fibronectin filaments {a-d, X480).

a-actinin suggests that, in addition to vinculin, a-ac-tinin may also be important in the formation of adhe-sion junctions during epithelial repair.21 Wound-hea-ling fibroblasts on day 14 showed a marked increasein anti-o:^ integrin staining, appearing similar tothat observed for nonmuscle myosin and a-actinin(Figs. 7C-D, arrows). Again, wound contraction re-sulted in the organization of a5(3{ integrin distributionto a lamellar pattern with reduced intensity similar tothat of a-actinin (Figs. 7E-F, arrows). Because en face

sectioning of radial wounds was required to identifystress fibers by transmission electron microscopy, weevaluated the distribution of a5j3, integrin in en facecryosections and compared its distribution with thatof f-actin and nonmuscle myosin (Fig. 8). Laser con-focal microscopy was used to eliminate out-of-focusblurring of fluorescence signals from above and belowthe plane of focus.

Evaluation of the 14-day wounds showed promi-nent phallacidin staining of f-actin bundles distrib-


No. 12 CORNEAL MYOFIDROBLAST / Gorono er ol 3277

Fig. 4. Anti-nonmuscle myosin staining of radial keratotomywounds 3 (a, b), 14 (c, d), and 30 (e, f) days after radial keratotomy.At day 3 anti-nonmuscle myosin staining appears increased alongthe margins of the wound (a, b, arrows). By day 14, fibroblasts thathave replaced the epithelial plug show increased staining, distrib-uted along the cell cortex (c, d, arrows). Regions along the cell ap-pear to show more intense staining in a linear fashion comparedwith other areas within the same cell (d, arrow). By day 30, nonmus-cle myosin staining appears organized in a lamellar pattern similarto that seen for f-actin and fibronectin (e, f, arrows) (original magni-fications: a, c, e: XI50; b, d, f; X375).

Fig. 5. Distribution of a-actinin at 3 (a, b), 14 (c, d), and 30 (c, f)days after radial keratotomy. At day 3 antibodies to a-actinin stainintensely the corneal epithelium (a, arrow) with more intense punc-tate staining along the basal epithelial cells (b, arrow). Adjacentkeratocytes show weak staining. By day 14, fibroblasts within thewound show prominent staining by antibodies to «-actinin (c, d,arrows). By day 30. a-actinin appears distributed along lamellarplanes similar to that seen for nonmuscle myosin (c, f, arrows)(original magnifications: a, c, e; XI50; b, d, f: X375).


0278 INVESTIGATIVE OPHTHALMOLOGY G VISUAL SCIENCE / November 1992 Vol. 33

Fig. 6. Colocalization of <x-actinin and nonmuscle myosin in a 14-day radial keratotomy wound. Antibodies to cv-actinin were detected usingrhodamine-conjugated goat anti-mouse IgG (a), whereas antibodies to nonmuscle myosin were detected using FITC-conjugated goat anti-rab-bit IgG (b). Note that the fibroblasts within the wound show an identical fluorescence distribution of nonmuscle myosin and «-actinin(arrows). On the other hand, the overlying epithelium (Epi) showed prominent a-actinin staining in the basal plasma membrane region,whereas nonmuscle myosin staining remained diffuse (curved arrows) (a, b, X900).

uted throughout the area of fibrosis (Figs. 8A-B,arrows). Sections stained with antinonmuscle myosinshowed a similar distribution (Fig. 8C). However, an-timyosin staining appeared to decorate the stressfibers in a beaded fashion (Fig. 8D, arrows) which wasconsistent with the known distribution of myosin instress fibers from fibroblasts in culture.22 The distribu-tion of «s/3| integrin at low magnification clearly indi-cated that integrin was not uniformly distributedthroughout the cell membrane but was localized tomultiple discrete punctate regions over the cell (Fig.8E). At higher magnification, areas of punctate stain-ing appeared as linear arrays, suggesting a stress fibercodistribution (Fig. 8F). Although reaction with otherj8i heterodimers cannot be ruled out, the distributionappears to be consistent with that of the fibronectinreceptor.

Discussion

The results from this study indicate that the initia-tion of wound contraction in nonpenetrating cornealwounds first involves the replacement of the epithelialplug by wound-healing fibroblasts, followed by the de-velopment of prominent intracellular stress fiberswithin the fibroblasts, which contain abundant f-ac-tin, nonmuscle myosin, and a-actinin. Because thedevelopment of stress fibers is one of the major char-acteristics associated with skin myofibroblasts,2'3

these results suggest that healing of RK-producedwounds involves the transformation of tissue kerato-

cytes into myofibroblast-like cells which are then re-sponsible for the development of corneal wound con-traction. Previous ultrastructural studies of cornealwound healing after RK have either not reported thepresence of myofibroblasts1114 or have suggested thatmyofibroblasts are not involved in corneal woundhealing.15 The failure of these earlier studies to iden-tify ultrastructural features of myofibroblasts in cor-neal wounds may be related, in part, to samplingerror. As observed in the current study, detection ofstress fibers required en face sectioning of radialwounds parallel to the wound surface. In standardcross sections of corneal tissue, however, microfila-ment bundles were not observed. These data suggestthat stress fibers in the cornea are uniquely orientedwithin the wound tissue such that conventional histo-logic sections of the wound provides only a cross-sec-tional view of the microfilament bundles, thus ex-plaining the inability to detect individual stress fibers.

Recently, Welch et al23 developed a hypotheticmodel for wound fibrosis and wound contraction inwhich the temporal relationship between extracellularfibronectin, f-actin organization, as@i integrin expres-sion, and wound contraction in a cutaneous woundmodel was described. It was suggested that fibronec-tin, which is synthesized by the fibroblasts, actsthrough the a5/3, integrin receptor to promote cell-cell and cell-substrate adhesion and the formation ofdense bundles of f-actin within the fibroblasts. Ac-cording to this model, wound fibroblasts undergo aseries of phenotypic modulations in which they elon-


No. 12 CORNEAL MYOFIDRODLAST / Gorono er ol 3279

Fig. 7. Distribution of a5j8, integrin at 3 (a, b), 14 (c, d), and 30 (e,f) days after radial keratotomy. At day 3 antibodies to a$fi, stain themargin of the wound localized predominantly to the basal plasmamembrane region of the migrating epithelium (a, b, arrows). By day14, fibroblasts within the wound show prominent staining by anti-bodies to a5#, (c, d, arrows). By day 30, a^t appears distributedalong lamellar planes similar to that seen for nonmuscle myosin (e,f, arrows) (original magnifications: a, c, e: X150; b, d, f: X375).

gate and extend bidirectionally, forming f-actin bun-dles along the lines of contractile force, pulling innewly deposited connective tissue around themselves.Because newly synthesized matrix is cross linked to

the surrounding tissue, the pulling in of matrixaround the cell results in the generation of tensionwithin the wound.

In the current study, the colocalization of f-actinand fibronectin in the cornea appear to support themechanism proposed by Welch et al23 for the skin.During corneal wound healing in the cat, intracellularstress fibers colocalize with extracellular fibronectinfilaments, which appear to share the same orientationand direction. These data suggest that the formationof intracellular stress fibers involves the pulling in andorganization of extracellular fibronectin to form fila-ments organized along the axis of the stress fibers.Because fibronectin has multiple matrix-binding do-mains, the movement of fibronectin along the cellsurface would likewise pull on other extracellular ma-trix molecules to which it is bound.

A critical link in the organization of intracellularf-actin and extracellular fibronectin is the integrin-fi-bronectin receptor. Integrin comprises a large familyof membrane-adhesion receptors of which the /?, fam-ily includes receptors for fibronectin, collagen, lami-nin, and human T-cell very late antigen hetero-dimers.24-25 The high-affinity integrin-fibronectin re-ceptor is a jS,. It has been demonstrated both in vivoand in vitro that integrin 0, subunit expression is partof the cellular adhesion plaque which is essential forproper assembly of fibronectin fibrils and that thisinteraction initiates the cytoskeletal events necessaryfor cell adhesion and spreading.18-24"26 In the currentstudy, it was not possible to identify the direct coloca-lization of a5^i integrin with f-actin or fibronectin be-cause of the weak fluorescence signal provided byanti-a5jff[ integrin and the difficulty in ruling out anycross over of the rhodamine signal into the FITCchannel. However, integrin fluorescence staining ap-peared to be more intensely localized along linear re-gions of the cell, which is consistent with a stressfiber-fibronectin colocalization for the a5/?, integrin-fibronectin receptor. Furthermore, myofibroblastscut in en face sections showed punctate areas of fluo-rescence, which is also consistent with the formationof adhesion plaques between the myofibroblast andthe extracellular matrix and would also link f-actin tothe fibronectin fibril through the putative fibronectinreceptor.

Overall, the fluorescence data indicate that, aftercorneal injury, keratocytes adjacent to the wound un-dergo a series of phenotypic alterations that first in-clude the expression of a5fil integrin, fibronectin,and a-actinin. This initial phenotypic change may oc-cur in response to the release of specific wound fac-tors, including serum fibronectin,27 platelet-derivedgrowth factor,28 and/or transforming growth factor-beta.29 As fibroblasts migrate into the wound and re-



Fig. 8. Distribution of f-actin (a, b), nonmuscle myosin (c, d), and «S|6| (e, f), 14 days after radial keratotomy. Tissue blocks were cut en faceand sections viewed on a BioRad MRC 600 laser scanning confocal microscope. Actin and nonmuscle myosin appear to be organized intobundles within cells similar to stress fibers seen in tissue culture cells (a-d, arrows). Integrin appears to have a punctate distribution withinwound-healing fibroblasts (e, arrow), which is organized into linear arrays similar to the stress fiber bundles (f) (a; c, e: X2I0; b, d, f: X650).

place the epithelial plug, interactions between extra-cellular fibronectin and a5(3{ integrin result in the poly-merization of actin and the formation of stress fiberswhich contain myosin and a-actinin. It is the forma-tion of stress fibers that is correlated with the initia-tion of wound contraction, suggesting a direct cause-and-effect relationship (already noted by Welch etal23). A further important characteristic of cornealwound contraction appears to be the redistribution ofextracellular fibronectin, intracellular actin, and ac-

tin-binding proteins from a randomly distributed pat-tern to a uniform distribution organized parallel tothe wound surface. Wound contraction does not ap-pear to involve simply a collapse of the extracellularmatrix surrounding the cell but would appear to in-volve an organized redistribution of matrix into a spe-cific pattern running parallel to the corneal surface.

How stress fibers generate tension across the woundand lead to tissue organization remains to be estab-lished clearly. In corneal wounds, the formation of


No. 12 CORNEAL MYOFIBROBLAST / Gorono er ol 3281

stress fibers appears to be related to the formation offibronectin filaments that run parallel with the f-actinfilaments. These data are consistent with the hypothe-sis that the formation of a stress fiber pulls togetherextracellular fibronectin to form fibronectin filamentswhich generates tension within the wound as de-scribed by Welch et al.23 Alternatively, stress fibersmay generate tension through a muscle-contractilemechanism in which fibers have an insertion and ori-gin directed along the major lines of tension withinthe wound. In the research reported by Welch et al,23

f-actin bundles appeared to be organized along thelines of contractile force which would support the al-ternative explanation for wound contraction. In thecornea, the contractile force would occur predomi-nantly across the wound. As noted, ultrastructuraland immunofluorescence evaluation of tissue cut incross and en face orientation indicate that microfila-ment bundles are organized predominantly parallel tothe wound surface and not necessarily across thewound. Furthermore, recent work by Petroll et al30

indicate that the three-dimensional orientation ofstress fibers in contracting corneal wounds becomesoriented along the wound rather than across thewound. Such findings do not support the mechanismas directly proposed by Welch et al.23 Further study ofthe stress fiber three-dimensional organization and itsrelationship to extracellular matrix proteins and thelines of contraction may provide important insightsinto the mechanism of wound contraction.

Finally, the apparent de novo expression of a-ac-tinin in corneal myofibroblasts suggests that this ac-tin-binding protein may be uniquely expressed bythese cells. Recent studies of a-actinin suggest thatthis actin-binding protein may be critically importantin both the formation of stress fibers and the bindingof f-actin filaments to the integrin receptor.31 Re-search by Pavalko and Burridge31 indicates that a-ac-tinin has an integrin-binding domain that, whenblocked or absent, results in the breakdown of theintracellular stress fibers. The finding that a-actinindoes not appear to be expressed in corneal kerato-cytes, although it is present in corneal myofibroblasts,suggests that the expression of a-actinin may be criti-cal to the initiation of wound healing and wound con-traction. Other muscle-specific proteins in addition toa-actinin have been identified in myofibroblasts fromother tissues, including the smooth muscle-specific a-isoform of actin and desmin.32 Although rabbit cor-neal myofibroblasts did not appear to contain muscleactins,13 the finding of a-actinin suggests that an im-portant phenotypic change between keratocytes andcorneal myofibroblasts is the expression of certainmuscle-specific genes. Further studies evaluating theregulation of expression of unique myofibroblast

genes may provide important insights into the mecha-nism of myofibroblast transformation and the woundcontraction mechanism.

In conclusion, contraction of corneal RK-producedwounds, as identified in earlier in vivo studies, is re-lated to the transformation of adjacent tissue kerato-cytes to a myofibroblast-like cell. Because parallelstudies reported elsewhere133 suggest that woundgape, as modulated by wound contraction, determinethe final refractive effect of RK, further studies of themyofibroblast, which lies at the heart of the mecha-nisms of wound contraction, may be crucial to theestablishment of predictability and permanencyfor RK.

Key words: actin, alpha-actinin, cornea, fibronectin, inte-grin, myofibroblast, myosin, radial keratotomy, refractivesurgery, wound healing

References

1. Jester JV, Petroll WM, FengW, Essepian J, andCavanagh HD:Radial keratotomy: I. The wound healing process and measure-ment of incisional gape in two animal models using in vivoconfocal microscopy. Invest Ophthalmol Vis Sci 33:3255,1992.

2. Majno G, Gabbiani G, Hirschel BJ, Ryan GB, and Statkov PR:Contraction of granulation tissue in vitro: Similarity to smoothmuscle. Science 173:548, 1971.

3. Gabbiani G, Hirschel BJ, Ryan GB, Statkov PR, and Majno G:Granulation tissue as a contractile organ: A study of structureand function. J Exp Med 135:719, 1972.

4. Gabbiani G, Ryan GB, and Majno G: Presence of modifiedfibroblasts in granulation tissue and their possible role inwound contraction. Experientia 27:549, 1971.

5. Bhathal PS: The presence of modified fibroblasts in cirrhoticlivers in man. Pathology 4:139, 1972.

6. Rudolph R, McLure WJ, and Woodward M: Contractile fibro-blasts in chronic alcoholic cirrhosis. Gastroenterology 76:704,1979.

7. Nagle RB, Evans LW, and Reynolds DG: Contractility of renalcortex following complete urethral obstruction. Proc Soc ExpBiolMed 148:611, 1975.

8. Nagle RB, Kneiser MR, Bulger RE, and Benditt EP: Inductionof smooth muscle characteristics in renal interstitial fibroblastsduring obstructive nephropathy. Lab Invest 29:422, 1973.

9. Skalli O, Schurch W, Seemayer T, Lagace R, Montandon D,Pittet B, and Gabbiani G: Myofibroblasts from diverse patho-logic conditions are heterogeneous in their content of actinisoforms and intermediate filament proteins. Lab Invest60:275, 1989.

10. Skalli O and Gabbiani G: The biology of the myofibroblastrelationship to wound contraction and fibrocontractive dis-eases. In The Molecular and Cellular Biology of Wound Re-pair, Clark RAF and Henson PM, editors. New York, Plenum,1988, pp. 373-402.

11. Jester J, Steel D, Salz J, Miyashiro J, Rife L, Schanzlin D, andSmith R: Radial keratotomy in non-human primate eyes. Am JOphthalmol 92:153, 1981.

12. Luttrull J, Smith R, and Jester J: In vitro contractility of avascu-lar corneal wounds in rabbit eyes. Invest Ophthalmol Vis Sci26:1449, 1985.

13. Jester JV, Rodrigues MM, and Herman IM: Characterization



of avascular corneal wound healing fibroblasts: New insightsinto the myofibroblast. Am J Pathol 127:140, 1987.

14. Yamaguchi T, Kaufman HE, Fukushima A, Safir A, and As-bell PA: Histologic and electron microscopic assessment of en-dothelial damage produced by anterior radial keratotomy inthe monkey cornea. Am J Ophthalmol 92:313, 1981.

15. Schachar RA, Black TD, and Huang T: Understanding RadialKeratotomy. Denison, TX, LAL Publishing, 1981, pp. 79-99.

16. Herman IM, Crisona NJ, and Pollard TD: Relation betweencell activity and the distribution of cytoplasmic actin and myo-sin. J Cell Biol 90:84, 1981.

17. RobertsCJ, BirkenmeierTM, McQuillan JJ, Akiyama SK, Ya-mada SS, Chen W-T, Yamda KM, and McDonald JA: Trans-forming growth factor beta stimulates the expression of fibro-nectin and of both subunits of the human fibronectin receptorby cultured human lung fibroblasts. J Biol Chem 263:4586,1988.

18. Akiyama SK, Yamada SS, Chen W-T, and Yamada KM: Anal-ysis of fibronectin receptor function with monoclonal antibod-ies: Roles in cell adhesion, migration, matrix assembly, andcytoskeletal organization. J Cell Biol 109:863, 1989.

19. Sanger JX, Sanger JM, and Jockusch BM: Differences in thestress fibers between fibroblasts and epithelial cells. J Cell Biol96:961, 1983.

20. Garana RMR, Herman I, Barry P, Cavanagh HD, and JesterJV: Confocal evaluation of the 3-dimensional cytoskeletal pro-tein organization in cat keratocytes. Invest Ophthalmol Vis Sci32(suppl):1143, 1991.

21. Zieske JD, Bukusoglu G, and Gipson IK: Enhancement of vin-culin synthesis by migrating stratified squamous epithelium. JCell Biol 109:571, 1989.

22. Langanger G, Moeremans M, Daneels G, Sobieszek A, Bra-bander MD, and Mey JD: The molecular organization of myo-sin in stress fibers of cultured cells. J Cell Biol 102:200, 1986.

23. Welch M, Odland G, and Clark R: Temporal relationships off-actin bundle formation, collagen and fibronectin matrix as-

sembly, and fibronectin receptor expression to wound contrac-tion. J Cell Biol 110:133, 1990.

24. Aldelda SM and Buck CA: Integrins and other cell adhesionmolecules. FASEB J 4:2868, 1990.

25. Buck CA and Horwitz AF: Integrin, a transmembrane glyco-protein complex mediating cell substratum adhesion. J Cell SciSuppl 8:231, 1987.

26. Mueller SC, Kelly T, Dai M, and Chen WT: Dynamic cytoskel-eton-integrin associations induced by cell binding to immobi-lized fibronectin. J Cell Biol 109:3455, 1989.

27. Postlethwaite AE, Keski-Oja J, Balian G, and Kang A: Induc-tion of fibroblast chemotaxis by fibronectin: Localization ofchemotaxis region to a 140,000 molecular weight nongelatinbinding fragment. J Exp Med 153:494, 1981.

28. Seppa HEJ, Grotendorst GR, Seppa E, Schiffman E, and Mar-tin GR: Platelet-derived growth factor is chemotactic for fibro-blasts. J Cell Biol 92:584, 1987.

29. Postlethwaite AE, Keski-Oja T, Moses HL, and Kang AH:Stimulation of the chemotactic migration of human fibroblastsby transforming growth factor-/?. J Exp Med 165:251, 1987.

30. Petroll WM, Cavanagh HD, Garana RMR, Barry PA, and Jes-ter JV: 3-Dimensional analysis of f-actin organization in myo-fibroblasts using laser confocal microscopy. ARVO Abstracts.Invest Ophthalmol Vis Sci 32(Suppl):l 163, 1991.

31. Pavalko KM and Burridge K: Disruption of the actin cytoskele-ton after microinjection of proteolytic fragments of «-actinin. JCell Biol 114:481, 1991.

32. Skalli O, Schurch W, Seemayer T, Lagace R, Montandon D,Pittet B, and Gabbiani G: Myofibroblasts from diverse patho-logic setting are heterogeneous in their content of actin iso-forms and intermediate filament proteins. Lab Invest 60:275,1990.

33. Petroll WM, New KC, Sachdev M, Cavanagh HD, and JesterJV: Radial keratotomy: III. Relationship between wound gapeand corneal curvature in primate eyes. Invest Ophthalmol VisSci 33:3283, 1992.


radial keratotomy. ii. role of the myofibroblast in...

Documents