Evidence for a Junctional Epithelial Attachment toCeramic Dental Implants*

A Transmission Electron Microscopic StudyRalph V. McKinney, Jr.,f David E. Steflikf and David L. Kothij:

Accepted for publication 18 March 1985

The interface of the crevicular gingiva with the surface of a dental implant is a criticalzone representing the potential biological seal which protects the underlying bone and softtissue-supporting mechanisms from destructive extraneous substances. Ultrastructural ex-amination of regenerated junctional epithelial cells interfacing surgically placed endostealdental implants, comprised of alpha-alumina oxide ceramic in single crystalline form,exhibited an external basal lamina and linear body located between the external surfaceepithelial cell and the implant. In addition, hemidesmosomes were located at intervals alongthe outer junctional epithelial plasma membrane. The component substructures of the basallamina and the hemidesmosomes were similar to those seen interfacing natural teeth. Thelinear body was an electron-dense structure between the lamina densa and the inertbiomaterial. This study provides ultrastructural evidence for the presence of an attachmentcomplex between gingiva and aluminum oxide implants which is analagous to that seenaround natural teeth. These data support the concept that a viable biological seal can developaround endosteal dental implants and provide support for satisfactory clinical service.

Modern endosteal dental implants have been usedfor approximately 45 years as a treatment for the re-placement of missing teeth in the partially or com-pletely edentulous patient. A critical feature that iscommon to all dental implant designs and materials isthe status of the soft tissue-dental implant interface.This transmucosal passage of the implant from theinternal aspects of the jaw bone and soft tissue into theexternal aspects of the oral cavity is a situation uniqueto dental implantology and makes the biological criteriafor dental implants different from implants used inorthopedic surgery. For the dental implant to survivein this dual environment, a viable biological seal thatprevents the ingress ofbacteria and other inflammation-producing agents must exist at the soft tissue-implantinterface.12

Examination of this biological seal is a difficult tech-nical problem because it requires manipulating soft andhard tissues simultaneously with implanted biomateri-

* This study was supported in part by Grant No. 7834-0001 fromKyocera, International, Inc. and Grant No. 1538-00-04 from Johnsonand Johnson, Inc.

t Department of Oral Pathology, Medical College of Georgia,Augusta, GA.

Department of Fixed Prosthodontics, University of North Car-olina at Chapel Hill, Chapel Hill, NC.

als. Methodology has improved considerably over re-cent years, and light microscopic examination of thegingival-biomaterial interface has become practical.3"6There are currently a number of reports detailing thehistological appearance of the gingival adaptation to aprotruding implant coronal post.7"11

A few workers have been able to prepare peri-im-plant-tissue specimens for examination by scanning andtransmission electron microscopy. McKinney and co-workers1213 have shown with scanning electron micros-copy that the regenerated gingiva forms a viable cuffaround the coronal implant post and extends epithelialpseudopodia from the junctional epithelium, similar tothat seen interfacing natural teeth, that contact thebiomaterial face. However, the scanning electron mi-croscope resolution prevents determining the exact na-ture of the cell organdes that are involved in theinterface biology between epithelial cells and the im-plant material.12

The first in vivo ultrastructural demonstration ofhemidesmosomes and a basal lamina between regen-erated junctional epithelia and a Vitallium dental im-plant was presented by James and Schultz14 using trans-mission electron microscopy. This was followed by thedemonstration of Listgarten and Lai of hemidesmo-somes and a basal lamina against epoxy resin dental

579

580 McKinney, Stefiik, Koth S. Periodontol.October, 1985endosseous implants in vivo.15 Most recently, othershave shown the presence of epithelial cell adhesionorgandes against titanium alloy (T-6A1-4V) and puretitanium in vitro16 and in v/vo.1718 Swope and James19found that these attachment organdies form as early as24 hours after Vitallium implant insertion.19

The purpose of this investigation was to examine,using transmission electron microscopy, the cells thatinterface with an endosseous dental implant comprisedof

-alumina oxide ceramic material. The evidence ofcellular activity and organdies similar to that seen inthe normal tooth-soft tissue interface would be strongevidence to support the suggestion that a viable biolog-ical seal exists around implanted ceramic materials andwould further denote the potential for long-term serv-iceability of these dental implants.

MATERIALS AND METHODSThe single-crystal sapphire dental implant is a rela-

tively new implant material introduced into the UnitedStates after initial investigative studies in Japan.2021The material consists of crystalline, -alumina oxideprepared in the configuration of a cylindrical screw(Fig. 1). The physical characteristics of the implantmaterial and its various designs have been reviewed.22

For this study, eighteen 20- to 25-kg mongrel dogshad single-crystal sapphire dental implants placed in anedentulous zone of the mandible that had been createdby the extraction of the third and fourth premolars 8weeks previously. The methods of animal preparation,implant surgical insertion and follow-up clinical eval-uation data have been covered in previous reports.1 K23'24Thirty-six single-crystal sapphire implants were placedbilaterally in the mandibles of these dogs.

The implants were left in place in the animal jaws

Figure 1. A photograph ofthe single crystal sapphire endosteal dentalimplant composed ofa-alumina oxide ceramic material. The implantsare available in several lengths and widths which allow for selectionofan appropriate sizefor the recipient host. The implant is extremelysmooth on its outer surface.

for periods up to 24 months. The dogs were clinicallyevaluated every 3 months using a standardized protocolwhich allowed statistical interpretation of the data.Results of these clinical trials and their statistical inter-pretation have been recently reported.24'25

At the beginning of the experiment, randomly se-lected animals were scheduled for sacrifice and micro-scopic examination. Some of the recovered specimenswere used for light microscopy, some for scanningelectron microscopy, and some for transmission elec-tron microscopy. Five animal specimens from 9, 12, 18and 24 months were selected for high resolution trans-mission electron microscopy (TEM).

Prior to TEM specimen preparation, the animalheads were fixed by vascular perfusion employing either3% phosphate buffered glutaraldehyde or 10% neutralbuffered formalin following a carotid artery cut-downprocedure. For perfusion, the fixative was administeredby a peristolic pump at 175 cc/minute at approximately80 mm of Hg for a period of 45 minutes. The externaljugular veins were severed to allow escape of the bloodand perfusate during the procedure.

Following this initial fixation, the mandibles contain-ing the implants fixed by 10% neutral buffered formalinwere block-resected, and the portion containing theimplant in situ was re-immersed in 10% formalin for48 hours. Upon completion of fixation, these blocksamples were washed in 0.1 m phosphate buffer (pH7.4) and dehydrated through a series ofgraded ethanolsof 50%, 70%, 95%, 95%, 100%, 100% for 12 hours ateach stage. The samples were then embedded in poly-methyl methacrylate (PMMA) according to our pub-lished procedure6 and were sectioned buccal-linguallywith a Buehler Isomet saw and diamond wafering bladeat either 120 or 2 to 5 mm.

The heads that were perfusion-fixed with 3% phos-phate buffered glutaraldehyde were processed differ-ently; following the initial 45-minute perfusion sched-ule, small 3x4 mm blocks of the free gingiva interfac-ing the implant were carefully microdissected awayfrom the implant and reimmersed in 3% glutaraldehydefor a total fixation time of 4 hours. The tissue dissectionwas accomplished by fine surgical instruments, magni-fying loops and careful attention to tissue orientation.Following the 4-hour fixation time, the gingival speci-mens were cubed to 1 mm3 for further TEM processing.Again, critical attention to tissue orientation was ob-served. The samples were washed three times (20 min-utes each wash) in 0.1 m phosphate buffer and post-fixed for 2 hours in 1 % osmium tetroxide buffered topH 7.4 with 0.1 m phosphate buffer. After post-fixation,the samples were again washed, then passed through agraded series of ethanols of 50%, 70%, 95%, 95%,100%, 100%, 100% for 10, 15, 20, 20, 30, 30, 30minutes, respectively. These samples were transferredthrough propylene oxide and embedded in Epon 812according to Luft's method.26

Volume 56Number 10 Junctional Epithelial Attachment to Ceramic Implants 581

The microdissected, Epon-embedded specimens werethin-sectioned using an LKB Ultratome III and glassknives. Orientation sections were cut at 1.0 , andthin sections were cut at approximately 75 nm.

Because of the concern that the microdissection pro-cedure may have disrupted structures external to theepithelial cell unit membrane, a cryofracture techniquewas developed using our PMMA-embedded specimens.For the cryofracture procedure, the 2 to 5 mm PMMAsections were frozen in liquid nitrogen for 30 secondsfollowed immediately by immersion in boiling distilledwater. This procedure resulted in the soft tissue fractur-ing away from the implant. The fracture plane occurredalong the implant-organic tissue interface because ofstructural differentials. The recovered cryofracturedgingival specimens were then reembedded in PMMAand polymerized. Orientation was simplified by the factthat the sections were previously histologically stainedthus providing anatomical markers.27

Thin sectioning was again accomplished with an LKBUltratome III with orientation sections stained in 1%toluidine blue and thin sections stained with uranylacetate and lead citrate.

All TEM sections were viewed with either an RCAEMU4C or a Philips 200 transmission electron micro-scope.

RESULTSOur transmission electron microscopic studies used

two methodologies to obtain high resolution informa-tion of the critical soft tissue-implant interface, a mi-crodissection technique and a cryofracture technique.To organize the results in a logical manner, we presentfirst the data obtained from microdissection, followedby the cryofracture technique.

Microdissected TEM Specimens. Low power electronmicroscopic views of the overall crevicular cell layerprovided orientation to the nature of the junctionalepithelium which was found to be 3 to 7 cells thick.The internal (or basal) layer ofjunctional epithelial cellsrevealed normal intercellular junctions, a prominentbasal lamina exhibiting both lamina lucida and laminadensa substructure, and anchoring fibrils which wereassociated with the underlying collagen fibers (Fig. 2).The external surface layer of junctional epithelial cellsdemonstrated normal cytoplasmic organelles: nucleus,tonofilaments, Golgi, ribosomes and endoplasmic retic-ulum (Figs. 2 and 3). The outermost plasma membranethat interfaced with the implant was primarily flat, butdid contain at various intervals cytoplasmic pseudopo-dia (Fig. 2a). The outer epithelial cells also showed thepresence of various flocculent densities in associationwith the membrane (Fig. 3). At various intervals alongthe outer cell surface, dense laminated organelles wereassociated with the plasma membrane (Figs. 3 and 4).These laminated organelles, or hemidesmosomes, were

present on flat portions of the cell membrane as well ason the membrane of pseudopodial projections emanat-ing from the cell (Fig. 5). Desmosomes were prominentbetween contacting intercellular bridges of the junc-tional epithelial cells (Fig. 2) and were similar in densityand structure to the hemidesmosomes noted on theouter plasma membranes.

Fine filaments extending from the body of the hem-idesmosome structure into the cytoplasm of the cellwere observed (Fig. 4). At very high resolution, thelaminated substructure of the hemidesmosome wasclearly revealed (Fig. 6).

The outermost cells of junctional epithelium con-tained a rich population of membrane-bound secretoryvesicles (Figs. 7 and 8). These vesicles were prominentthroughout the cell cytoplasm including associationwith Golgi (Fig. 9), but appeared more densely aggre-gated at or near the implant front unit membrane (Figs.7 and 8). Some of the vesicles gave the appearance ofcoalesced aggregates at the implant front (Fig. 7), andvesicles often were aligned with the outer plasma mem-brane (Fig. 8). Occasionally, fusion of the vesicularmembrane with the outer plasma membrane was ob-served suggesting an opening for the discharge of thevesicle contents externally (Fig. 10).

A varied population of vesicles was observed in thejunctional epithelium at high resolution: single unit,membrane-bound vesicles and double unit, membrane-bound vesicles (Fig. 11 ). Both of these type of vesicleswere found intimately related with endoplasmic retic-ulum and Golgi (Figs. 9 and 11 ). Very high magnifica-tion of other vesicles demonstrated a central densearrangement ofcontents surrounded by an outer lucentzone containing dense strands of material (Fig. 12).

Some sections revealed secretory vesicles that wereirregular in shape and contained electron dense floc-culent material of varying density that was similar inmorphology to material found on the outer unit mem-brane (Fig. 13). Tonofilaments were present in varyingnumbers in the junctional epithelium (Figs. 3, 7, 8 and13).

Cryofractured TEM Specimens. The cryofracturedjunctional epithelium revealed adequate morphologicalretention of cellular architecture with clearly distin-guishable nuclei, nuclear envelopes, nucleoli, intercel-lular bridges, desmosomes, ribosomes, endoplasmic re-ticulum and mitochondria in spite of the formalinfixation and double PMMA embedding procedures(Fig. 14). Some artifactual disruption of the cell sap wasevident at high magnification, but cell membranes andtheir attendant structures remained intact (Fig. 16).Ultrastructural examination of the most superficialjunctional epithelial cells interfacing the implant re-vealed a clearly distinguishable unit membrane, anexternal basal lamina with lamina lucida and laminadensa components, and clearly definable hemidesmo-somes (Figs. 15, 16 and 17). The lamina densa and

582 McKinney, Stefiik, Koth J. Periodontol.October, 1985

Figure 2. Low power Iransmission electron micrographs ofthe regeneratedjunctional epithelium that was microdissectedfrom around a sapphireimplant, a. The external surface junctional epithelial cells interfacing the implant (space at top left contained implant) demonstrate intercellularbridges with desmosomes (d), intercellular spaces (*), nucleus (N), surface pseudopodial projections (arrowheads) and two polymorphonuclearleukocytes (P) which are frequently seen in junctional epilhelia. Present throughout the two cells are electron dense secretory vesicles (V), that areaggregated in greater numbers at the external unit membrane of the epithelial cell interfacing the implant (magnification x 8000). b. Basal cellsofthe junctional epithelium exhibit intercellular bridges (arrowheads), desmosomes (d), intercellular spaces (*), nucleus (N) and the lamina lucida(11) and lamina densa (Id) components of the basal lamina. Anchoring fine filaments (ff) extend from the basal lamina into the supportingconnective tissue around collagen fibers (c). Hemidesmosomes (h) appear as dense plaques on the cell membrane interfacing the basal lamina(magnification x 18,000).

Figure 3. transmission electron micrograph ofan interfacing junc-lional epithelial cell in a microdissected gingival specimen. The im-plant occupied the space at the top left. Organelles are prominentthroughout the cell and include the nucleus (N), rough endoplasmicreliculum (er), free ribosomes and numerous tonoftlaments (tf). Notpresent in this particular cell are the numerous secretory vesicles seenin Figure 2a. The external plasma membrane exhibits a dense plaquestructure suggestive ofa hemidesmosome (h) and scattered flocculentdeposits (arrowheads) associated with the membrane (magnificationX 10,400).

Figure 4. A hemidesmosome (h), located on the external unit mem-brane of an interfacing epithelial cell exhibits a fine filamentousprocess emanating from the hemidesmosome into the cell cytoplasm(arrowheads). Note the variable density of the vesicle contents (V) inthis microdissectedgingival specimen. The cell cytoplasm and nucleus(N) are slightly out offocus since fine focusing was concentrated on

Figure 5. Three prominent hemidesmosomes (arrows) present on apseudopodial projection extending from an interfacing junctional ep-ithelial cell. External basal lamina structure is not visible in thissurgically dissected specimen (magnification x 38,750).

WBSKmMFigure 6. High resolution transmission electron micrograph revealingthe subunit structures of a hemidesmosome on the interfacing unitmembrane of a microdissected specimen. The implant occupied thespace at the figure top. Visible is the unit membrane (urn), peripheralbar or density (pd), pyramidalparticles (below dots) on unit membraneandfinefilaments (arrowheads) extendingfrom the hemidesmosome.An external basal lamina is not demonstrated (magnification x177,500).

the hemidesmosome, filamentous process and the vesicle. The implantwas located at the top of the picture (magnification x 23,400).

J. Periodontol.584 McKinney, Stefiik, Koth October, 1985

Figure 7. Transmission electron micrograph of microdissected junctional epithelial specimen removed from around the implant. Numeroussecretory vesicles (V) are aggregated at the implant front near the outer unit membrane (lop ofmicrograph) in the first tier ofepithelial cells (1).Fewer vesicles occur in the second tier (2) of epithelial cells. Desmosomes are present between epithelial projections (arrowheads) (magnificationx 18,900).

Figure 8. Transmission electron micrograph ofmicrodissectedjunctional epithelial specimen removedfrom around the implant. Secretory vesicles(V) containing dark material are aligned along the implant front ofan interfacing junctional epithelial cell. Two vesicles appear to have coalescedwith the unit membrane (arrows) (magnification x 25,200).

Volume 56Number 10

Figure 9. Transmission electron micrograph of microdissected func-tional epithelial specimen removedfrom around the implant. Secretoryvesicles of varying density, and rihosomes surround a Golgi (G) in ajunctional epithelial cell (magnification x 25,200).

Figure 10. Transmission electron micrograph ofmicrodissected junc-tional epithelial specimen removed from around the implant. Highresolution micrograph of a membrane-hound secretory vesicle fusedwith the epithelial plasma membrane at the implant front and reveal-ing an opening to the external environment. Vesicular content dis-charge is indicated by the presence ofsimilar dense material withinthe vesicle and extra cellularly (arrow) (magnification x 44,800).

Figure 11. Transmission electron micrograph ofmicrodissectedjunc-tional epithelial specimen removed from around the implant. Highresolution electron micrograph showing a double unit membranevesicle (double arrow) and a single membrane-bound vesicle (singlearrow) with contents of variable density, near a portion of Golgi (G)(magnification x 46,000).

lamina lucida varied in width along the implant frontwith the most variance being shown by the laminadensa (Figs. 15-18). The lamina densa varied from aneven width, electron-dense structure to a varied struc-

Junctional Epithelial Attachment to Ceramic Implants 585

Figure 12. Transmission electron micrograph ofmicrodissected junc-tional epithelial specimen removed from around the implant. Twoepithelial vesicles which exhibit dense center zones surrounded by adistinctive outer rim, all contained in a single unit membrane (mag-nification x 92,300).ture revealing an internal complex organization (Figs.17 and 18). At times the outer limit of the lamina densawas flocculent and slightly indistinct in its outline (Figs.17 and 18). Close examination of the lamina densasuggested the presence of two distinct electron-denselayers (Figs. 17 and 18). At very high resolution, anarrow electron-lucent band was observed between thelamina densa and the outermost electron-dense layer(Fig. 18).

High magnification electron microscopy also allowedresolution of the substructure of the hemidesmosomesto include the peripheral densities, pyramidal particlesand fine filaments associated with the unit membraneand basal lamina (Figs. 6 and 15-18).

DISCUSSIONDental implants have increased in sophistication

since their modest modern beginnings in the early1940s. The literature contains numerous articles con-cerning implant longevity, clinical serviceability, func-tion and even rudimentary statistical analyses ofhumancases.28'29 However, very little attention has been ac-corded the biological interface between hard and softtissues and the implanted material. The fact that im-plants do function for extended periods of time andprovide adequate and satisfactory service to many pa-tients raises the question as to what causes or createsan acceptable environment in the human oral cavity.

This investigation conclusively demonstrates that re-generated crevicular gingiva around ceramic dental im-plants develops junctional epithelial organelle and cel-lular architecture that is similar to the morphologicalstructures that interface the natural tooth.30"36 The pres-ence of these structures adjacent to the implant faceprovides biological evidence for successful long-termclinical service of the ceramic endosseous dental im-plants.

Secretory Vesicles. Our initial experimentation withthe microdissected specimens showed the presence ofjunctional epithelial cells and their organdes as de-scribed by Schroeder.32 We thus expected to find in the

586 McKinney, Stefiik, Roth J.Periodontol.

October, 1985

Figure 13. A junctionai epithelial cell dissectedfrom the implant (space, upper left) containing two vesicles (V) ofvarying shape and morphologicalcontents. The density of the flocculent material along the outer unit membrane (arrows) is similar to that observed in the vesicles, although noorganized basal lamina is visible along this microdissected specimen cell surface. Tonofilamenls are prominent in the cytoplasm (tf) (magnificationx 82,000).

Figure 14. The basal cell layer of junctionai epithelium from a formalin-fixed, PMMA-embedded specimen that was reprocessed for thecryofracture technique (see text). Note the adequate retention ofoverall cell morphology incuding nucleus (N), nucleolus (Nu), ribosomes (r), roughendoplasmic reticulum (er), intercellular bridges (arrows) and intact desmosomes (d) (magnification x 42,500).

Volume 56Number 10 Junctional Epithelial Attachment to Ceramic Implants 587

Figure 15. High resolution transmission electron micrograph ofthe outer junctional epithelial cell unit membrane. The unit membrane interfacingthe implant demonstrates periodic hemidesmosomes (h) composed ofperipheral densities and pyramidal particles (pd) with fine filaments (if).The basal lamina substructure oflamina lucida and lamina densa is very distinct, although the lamina densa varies in width (between arrowheads)as well as outer surface morphology. This specimen was formalin-fixed, PMMA-embedded then processed for the cryofracture technique andreembedded in PMMA. The implant occupied the top space in this figure (magnification x 89,700).

Figure 16. This section ofan epithelial cell membrane reveals a lamina lucida (11) wider than the lamina densa (Id). Hemidesmosomes (h) on theunit membrane demonstrate distinct peripheral densities with pyramidal particles (below dots). Fine filaments emanate from the density into thelamina lucida and cell cytoplasm. The nature of the surface dense particle is unknown. Processing artifact most likely is present at the asterisks.This specimen was formalin-fixed, PMMA-embedded then processed for the cryofracture technique and reembedded in PMMA. The implantoccupied the top space in this figure (magnification x 131,200).

junctional epithelium cells a normal complement oforgandes including endoplasmic reticulum, Golgi andnumerous tonofilaments. The one type of organdiethat was not expected in such great abundance was thesecretory vesicle. These vesicles were associated withGolgi, were numerous throughout the cytoplasm of thecell and were observed fused with the outer plasmamembrane suggesting a mechanism for discharge oftheir vacuole contents. Presence ofthese vesicles aroundthe Golgi and throughout the cytoplasm further sug-gests that they were produced by the cells in concertwith current accepted cell biology doctrine, i.e., proteinand other materials are produced by the endoplasmicreticulum and transported to the Golgi where the ma-terial is packaged in membrane-bound vesicles. Thevesicles then move through the cytoplasm to eventually

discharge their product via exocytosis to the externalenvironment. This type of process would seem to ac-count for the presence of the flocculent, dense materialon the outer cell membrane which is similar in densityto the material contained in the vesicles. Indeed, thismay partially explain how the basal lamina and otherextracellular proteoglycans on the external surface ofthe junctional epithelial cells are produced. The reasonthat there are several morphological populations ofvesicles is not clear. However, it is known that kerato-hyalin granules in oral mucosa, which are involved inorthokeratin or parakeratin differentiation activities, dohave varying morphological features suggesting differ-ent functions for the granules.34 It may be that thereare specific targeted reasons for each of these vesicles,one to produce the basal lamina, the other to produce

588 McKinney, Stefiik, Roth J. Periodontol.October, 1985

Figure 17. This epithelial cell exhibits numerous hemidesmosomes along the unit membrane with prominent peripheral densities (h). Thepyramidal particles andfine filaments are not as distinct as in Figures 15, 16 and 18. The lamina densa varies from an even outer surface (e) toa jlocculent uneven (u) surface. Note the complex organization of the lamina densa at both e and u. This specimen was formalin-fixed, PMMA-embedded then processedfor the cryofracture technique and reembedded in PMMA. The implant occupied the top space in this figure (magnification 110,700).

Figure 18. A junctionai epithelial cell unit membrane with long hemidesmosomes revealing their substructure ofperipheral densities and pyramidalparticles (below dots). The lamina lucida is narrower than previous figures. The complex organization of the lamina densa (Id) and the outerdense linear body (lb) is revealed (see text). A sublamina lucida occurs al s. This specimen was formalin-fixed, PMMA-embedded then processedfor the cryofracture technique and reembedded in PMMA. The implant occupied the top space in this figure (magnification x 170,00).

additional extracellular matrix components such as gly-cosaminoglycans or proteoglycans. Further study isneeded to clarify this point. Despite the unclear natureof the multiple vesicle population, the presence of alarge number of vesicles aggregated at the junctionaiepithelial surface very strongly supports the suppositionthat the epithelial cells produce their own basal laminaand other extracellular attachment components andthus are directly responsible for their survival in theoral environment.

Hemidesmosomes. The hemidesmosome organellewas of great interest because it is intimately involved inepithelial attachment to natural teeth30"32 and has beenobserved ultrastructurally by other implant scien-tists.141517"19 We initially observed hemidesmosomeson our microdissected specimens as dense, plaque-likeconfigurations on the outer unit membrane. Because

these latter specimens lacked a well-structured basallamina, we were concerned that we might not be posi-tively observing hemidesmosomes. When we employedhigh resolution microscopy of the microdissected spec-imens, we were able to show the unit structure of thehemidesmosomes complete with fine filaments runninginto the cell cytoplasm (Figs. 4 and 6).

Upon using the alternative cryofracture technique,we significantly improved our ability to preserve theintegrity and intimate detail of the external epithelialsurface interfacing the dental implant. Present were theinner and outer peripheral densities, pyramidal particlesand fine filaments of the hemidesmosome (Figs. 15 and16) as detailed by others.30 The fine filaments coursedthrough the inner peripheral density from the cell sapand extended through the outer peripheral density intothe lamina densa (Figs. 15 and 16).

Volume 56Number 10

Basal Lamina. Early ultrastructural examination ofthe microdissected specimens did not reveal a promi-nent basal lamina, only irregular flocculent densitiesalong the outer plasma membrane (Fig. 13). Instead ofaccepting this as lack of evidence for a basal laminainterfacing implants, we considered the possibility thatthe microdissection technique had torn the extracellularattachment structures or had created an artifact. Thus,we employed the cryofracture technique. However, theonly specimens remaining that could be used for thisprocedure were the formalin-fixed PMMA-embeddedspecimens. Fortunately, the formalin fixation was ade-quate to preserve satisfactory ultrastructural detail andthe basal lamina, plasma membrane and hemidesmo-somes were vividly displayed by this procedure (Figs.14-18). Use of cryofracture not only revealed newultrastructural details, but also demonstrated the neces-sity of employing more than one procedure in under-taking morphological studies.

The basal lamina appeared as a definitively layeredstructure on the outer junctional epithelial unit mem-brane. Based on the degree of resolution in the highmagnification transmission electron micrographs, thesubstructures of the basal lamina, the lamina lucidaand the lamina densa were evident. A sublamina lucida,as described by Kobayashi and co-workers,30 was for-tuitously observed in some high resolution micrographs(Fig. 18). The lamina densa had an uneven outer borderin some specimens and appeared to be composed oftwo electron-dense complex structures separated by athin electron-lucent layer, the sublamina lucida (Fig.18). The existence of this additional dense structureoutside the lamina densa could be analogous to thelinear border or dental cuticle as described by Ko-bayashi et al.30 Although the presence and function ofa dental cuticle is debated, the fact remains that thesedense structures have been ultrastructurally shown innatural teeth.3032 The cuticle is theorized to be thebiological layer between the tooth surface and the sub-lamina lucida of the basal lamina. The linear border isthe dense structure observed between afibrillar cemen-tum and the basal lamina in place of a dental cuticle.30James and Kellen4 earlier suggested that a dental cuticledid exist between implant and junctional epitheliumbased on their histological periodic acid-Schiff (PAS)staining studies. We do not agree with their observation,or the concept that a dental cuticle exists around im-plant posts, but we do agree that there conceivablyexists a linear border structure outside the lamina densathat is intimately involved with the attachment appa-ratus. The linear border is well defined and probablyrepresents the condensation ofextracellular matrix pro-teoglycans involved in the attachment of the basallamina to surrounding structures. Since filaments orprotein structures cannot insert into inert biomaterialsas they do into cementum, the linear border in ourview is a very important structure that may representcondensation of proteoglycans, possibly fibronectin,

Junctional Epithelial Attachment to Ceramic Implants 589that attaches the lamina densa to a coating of glycosa-minoglycans on the surface of the ceramic biomaterial.

The Attachment Complex. The ultimate questionremains as to what is the nature of the molecularadhesion of the junctional epithelium to the implant.Although this question is not completely answered inthis study, the ultrastructural data presented can becombined with some current concepts of cell biologyto achieve a better understanding of the implant-tissue-interface attachment complex.

Laminin is an extracellular glycoprotein that is lo-cated in the lamina lucida and is probably responsiblefor the adhesion of the epithelial cells to the collagenousbasal lamina.37 38 It is likely produced by the epithelialcells themselves,39 and this would agree with our obser-vations of the material contained in secretory vesiclesbeing similar in density to material attached to theplasma membrane.

The lamina densa is devoid of laminin but has pres-ent another glycoprotein involved in adhesion activi-ties, fibronectin, a glycoprotein produced by fibroblastsand endothelial cells and found in the extracellularmatrix, especially during wound healing.40 Fibronectinbinds to collagen and glycosaminoglycans,41 and thusmay represent the "glue" between the implant bioma-terial and the basal lamina's lamina densa which iscomposed of Type IV collagen.42

The glycosaminoglycans, hyaluronic acid, heparinsulfate and heparin (called proteoglycans when com-plexed to a protein core), are produced by fibroblastsduring the healing phase following implant insertionand may coat the implant surface. Fibronectin, likewiseproduced by these same fibroblasts, is present in theextracellular matrix. As the junctional epithelium re-generates, producing its own basal lamina and laminin,the fibronectin provides the bond between the proteo-glycans (or glycosaminoglycans) on the implant surfaceand the lamina densa of the basal lamina. Thus, thedense outer layer, or linear body of Kobayashi et al.,30seen in our study may be the electron-dense staining ofthe glycoprotein fibronectin and may portray the ulti-mate ultrastructural biological seal.

The hemidesmosomes are undoubtedly epithelial-specific, adhesion plaques43 that "tack" the plasmamembrane to adjacent structures or substrate. Interest-ingly, the adhesion plaques, and presumably the hemi-desmosomes, do not contain fibronectin,44 althoughanother actin-binding protein, vinculin, is present.43

Conclusions. This study, using ceramic alumina ox-ide endosteal implants, demonstrates the ultrastructuralpresence of anatomical structures that represent thebiological attachment ofgingival epithelium to the den-tal implant. Present are hemidesmosomes, basaflaminaand a linear border on the external unit membrane ofthe epithelial cells. These structures are similar in natureto those observed in mature31,32 and regenerated35 36junctional epithelium adjacent to a tooth. The presenceof these structures provides tangible biological evidence

590 McKinney, Stefiik, Koth i. Periodontol.October. 1985for the presence of a per-perimucosal or biological seal2at the tissue-implant interface and validates currentclinical studies demonstrating satisfactory patient serv-ice with ceramic implants.45'46

REFERENCES

1. Lavelle, C. L. B.: Mucosal seal around endosseous dental im-plants. J Oral Implantai 9: 357, 1981.

2. McKinney, R. V., Jr., Stefiik, D. E., and Koth, D. L.: Per, perior trans? A concept for improved dental terminology. J Proslhet Dent52:267, 1984.

3. Klawitter, J. J., and Hulbert, S. F.: Application of porousceramics for the attachment of load bearing internal orthopedicapplications. J Biomed Mater Res, Symposium No. 2(Part I). 161,1971.

4. James. R. ., and Kellen, E. E.: A histopathological study ofthe nature of the epithelium surrounding implant posts. Oral Implan-tla: 105 and 137, 1973.

5. Gross, U. M., and Strunz, V.: Surface staining of sawed sectionsof undecalcified bone containing alloplastic implants. Satin Technol52: 217, 1977.

6. Stefiik, D. E.. McKinney, R. V., Jr., Mobley, G., and Koth, D.L.: Simultaneous histological preparation of bone, soft tissue andimplanted biomaterials for light microscopic observations. StainTechnol 57: 91, 1982.

7. Bodine, R. L., and Mohammed, C. I.: Histologie studies of ahuman mandible supporting an implant denture. J Prosthet Dent 21:203, 1969.

8. Natiella, J. R., Armitage, J. E., Meenaghan. . ., and Greene,G. W.: Tissue response to dental implants protruding through mucousmembrane. Oral Sci Rev 5: 85, 1974.

9. James, R. ., and Kellen, E. E.: A histological report on thenature of the epithelium and underlying connective tissue whichsurrounds oral implants. II. J Biomed Mater Res, Symposium No.5(Part 2), 373, 1974.

10. Brunski. J. B., Moccia, A. F., Jr., Pollack, S. R., et al.: Theinfluence of functional use of endosseous dental implants on thetissue-implant interface. I. Histological aspects. J Dent Res 58: 1953,1979.

11. McKinney, R. V., Jr., and Koth, D. L.: The single crystalsapphire endosteal dental implant: material characteristics and 18-month experimental animals trials. J Prosthet Dent 47: 69, 1982.

12. McKinney, R. V., Jr., Stefiik, D. E., and Koth, D. L.: Thebiologic response to the single-crystal sapphire endosteal dental im-plant: scanning electron microscopic observations. J Prosthet Dent51: 372, 1984.

13. McKinney, R. V., Jr. Stefiik, D. E., and Koth, D. L.: Ultra-structural surface topography of the single crystal sapphire endosseousdental implant. J Oral Implantol 11: 327, 1984.

14. James, R. ., and Schultz, R. L.: Hemidesmosomes and theadhesion of junctionai epithelial cells to metal implants. A prelimi-nary report. J Oral Implantol 4: 294, 1974.

15. Listgarten, . ., and Lai, C. H.: Ultrastructure of the intactinterface between an endosseous epoxy resin dental implant and thehost tissues. J Biol Buccale 3: 13, 1975.

16. Gould, T. R. L., Brunette, D. M., and Westbury, L.: Theattachment mechanism of epithelial cells to titanium in vitro. JPeriodont Res 16: 611, 1981.

17. Karagianes, M. T., Westerman, R. E., Hamilton, A. I., et al.:Investigation of long term performance of porous-metal dental im-plants in nonhuman primates. J Oral Implantol 10: 189, 1982.

18. Gould, T. R. L.. Westbury, L.. and Brunette, D. M.. Ultra-structural study of the attachment of human gingiva to titanium invivo. J Proslhet Dent 52: 418, 1984.

19. Swope, E. M., and James, R. .: A longitudinal study on

hemidesmosome formation at the dental implant-tissue interface. /Oral Implantol 9: 412, 1981.

20. Kawahara, H.: Today and tomorrow of bioceramics in artifi-cial organ. Jpn Art ifOrgan 6: 218, 1977.

21. Kawahara, H., and Hirabagashi, M.: Single crystal aluminafor dental implants and bone screws. J Biomed Mater Res 14: 597,1980.

22. McKinney, R. V., Jr., Koth, D. L., and Stefiik, D. E.: Thesingle crystal-sapphire endosseous dental implant. I. Material char-acteristics and placement techniques. J Oral Implantol 10:487, 1982.

23. Koth, D. L., and McKinney, R. V., Jr.: The single crystalsapphire endosteal dental implant. J. W. Clark (ed), Clinical Den-tistry, Chapter 62, vol 5. Philadelphia, Harper and Row, 1984.

24. McKinney, R. V., Jr., Koth, D. L., and Stefiik, D. E.: Thesingle crystal sapphire dental implant. II. Two year results of clinicalanimal trials. J Oral Implantol 10: 619, 1983.

25. Stefiik, D. E., McKinney, R. V., Jr., and Koth, D. L.: Astatistical analysis of the clinical response to the single-crystal sapphireendosseous dental implant in dog jaws. J Dent Res 62: 1212, 1983.

26. Luft, J. H.: Improvements in epoxy resin embedding methods.J Biophys Biochem Cytol 9: 409, 1961.

27. Stefiik, D. E., McKinney, R. V., Jr., Koth, D. L., and Singh,B. B.: The biomaterial tissue-interface: a morphological study utilizingconventional and alternative ultrastructural modalities. Scan ElectronMicrosc/II: 547, 1984.

28. Linkow, L. I.: Statistical analysis of 173 implant patients. OralImplantol 4: 540, 1973.

29. Cranin, . ., Rabkin, M. F., and Garfinkel, L.: A statisticalevaluation of 952 endosteal implants in humans. J Am Dent Assoc94: 315, 1977.

30. Kobayashi, K, Rose, G. G., and Mahan, C. J.: Ultrastructureof the dento-epithelial junction. J Periodont Res 11: 313, 1976.

31. Listgarten, . .: Electron microscopic study of the gingivo-dental junction of man. Am JAnat 119: 147, 1966.

32. Schroeder, H. E.: Ultrastructure of the junctionai epitheliumof the human gingiva. Helv Odontol Acta 13: 65, 1969.

33. Yamaski, ., Nikai, H., Niilani, K, and Ijuhin, N: Ultrastruc-

ture of the junctionai epithelium ofgermfree rat gingiva. J Periodontol50: 641, 1979.

34. Jessen, H.: Two types of keratohyalin granules. / UltrastructRes 33: 95, 1970.

35. Taylor, A. C, and Campbell, M. M.: Reattachment of gingivalepithelium to the tooth. J Periodontol 43: 281, 1972.

36. Braga, A. M., and Squier, C. .: Ultrastructure of regeneratingjunctionai epithelium in the monkey. J Periodontol 51: 386, 1980.

37. Terranova, V. P., Rohrbach, D. ., and Martin, G. R.: Role

of laminin in the attachment of PAM 212 (epithelial) cells to base-ment membrane collagen. Cell 22: 719, 1980.

38. Farquhar, M. G.: The glomerular basement membrane. E. D.Hay (ed), Cell Biology of Extracellular Matrix, pp 335-378. NewYork, Plenum, 1981.

39. Hogan, B. L. M., Cooper, A. R., and Kurkinen, M.: Incorpo-ration into Reichert's membrane of laminin-like extracellular proteinssynthesized by parietal endoderm cells of the mouse embryo. DevBiol SO: 289, 1980.

40. Hakomori, S., Fukuda, M., Sekiguchi, K., and Carter, W. G.:Fibronectin, laminin, and other extracellular glycoproteins. K. A.Piez and A. H. Reddi (eds). Extracellular Matrix Biochemistry, pp229-275. New York, Elsevier, 1984.

41. Oldberg, ., and Ruoslahti, E.: Interactions between chondro-itin sulfate proteoglycan, fibronectin, and collagen. J Biol Chem 257:4859, 1982.

42. Timpl, R., Wiedemann, H., van Delden, V., et al.: A networkmodel for the organization of type IV collagen molecules in basementmembranes. Eur J Biochem 120: 203, 1981.

43. Geiger, .: A 130K protein from chicken gizzard: its locali-zation at the termini of microfilament bundles in cultured chickencells. Cell 18: 193, 1979.

Volume 56Number 10

44. Chen, W. T., and Singer, S. J.: Fibronectin is not present inthe focal adhesions formed between normal cultured fibroblasts andtheir substrata. Proc Nati Acad Sci USA 77: 7318, 1980.

45. Koth, D. L., McKinney, R. V., Jr., and Davis, Q.: The single-crystal sapphire endosteal dental implant. A longitudinal humanstudy, one year results. J Prosthet Dent 50: 72, 1983.

46. Stefiik, D. E., Koth, D. L., and McKinney, R. V., Jr.: A

Junctional Epithelial Attachment to Ceramic Implants 591clinical and statistical analysis of human clinical trials with the singlecrystal sapphire endosteal dental implant: two year results. J OralImplanto! 11: 500, 1984.

Send reprint requests to: Dr. Ralph V. McKinney, Jr., Departmentof Oral Pathology, School of Dentistry, Medical College of Georgia,Augusta, GA 30912.