Annals of the Royal College of Surgeons of England (I976) vol 58

Keratinization of the oral epithelium

David Adams BSC MDS PhD

Department of Oral Biology, Welsh National School of Medicine Dental School, Cardiff

SummaryThe morphology of the keratinizing epi-thelia in the mouth is reviewed in the light ofrecent knowledge. There appears to be a spec-trum of degrees of keratinization rather thandistinct types, and the degree of keratinizationis reflected in the degree of packing and orien-tation of tonofilaments. The role of keratohya-line and other granules in the process is dis-cussed and it is suggested that modificationof the cell membrane is an important part ofkeratinization. Although the potential of thevarious areas in the mucosa is genetically de-termined and appears early in fetal life, theconnective tissue exerts an influence on theextent of keratinization of the surface in amanner which is not understood.

IntroductionThroughout the animal kingdom epithelialtissues play very varied roles in the protectionof the underlying tissues. Hair, feather, horn,claw, hoof, and wool have as a common in-gredient in their make-up the keratins, a macro-molecular, resistant, insoluble protein class.The process by which the protein is madetough and insoluble is known as keratinization,a process that results in a wide range of prod-ucts and whose nature is far from being fullyunderstood. In the oral cavity the lining mu-cous membrane becomes keratinized to vary-ing degrees in different animals and also indifferent areas of the mouth.

In man, in contrast to the common labora-torv animals, the epithelium of the mouth has asurface which varies in its degree of keratiniza-tion from none at all, as in the cheek, throughpartial or parakeratinization, as found some-times on the gingiva and parts of the palate,to full or orthokeratinization, as seen in mostparts of the hard palate and overlying the gin-giva where it is attached to bone.

In this review I shall consider the morpho-logy of the oral keratinizing epithelium and

compare it with the non-keratinizing mucosaand the epidermis. I also wish to examinesome of the mechanisms which control andregulate keratinization and discuss briefly theclinical implications of these.

Orthokeratinization, in which the surfaceundergoes cornification as cells lose their stain-ing characteristics and their nuclei, is found onthe hard palate and on gingiva, especiallywhere this is firmly bound down tounderlyinlg bone (Fig. i). The epitheliumlies on a basement membrane which separatesit from the connective tissue. Above this thereis a series of more or less well-defined layers.First comes the germinal layer, one or twocells thick, then the stratum spinosum, withits characteristic prickles, after which thestratum granulosum may appear, and finallythe stratum corneum, which varies from a verythin layer perhaps only two cells thick to one

FIG. I Section of stratified squamous epi-thelium fom the palate showing orthokeratin-ization. Haematoxylin and eosin.

Arnott Demonstration presented OI1 2nd December 1975

352 David Adams

of io or more cells in thickness. I will exam-ine each of these in turn in the light of recentknowledge.

Basement membraneThis structure can be visualized with the lightmicroscope, particularly after staining with theperiodic-acid-Schiff (PAS) technique. Therehas been confusion over the relationship of thelight microscope appearance to the structuresseen in this area with the electron microscope(Fig. 2). The dense layer seen in electron micro-scope preparations is too narrow (of the orderof 50 nm (5ooA)) to be the structure seen withthe light microscope (which has a dimensionof approximately i ,um). The electron-denselayer or lamina densa is separated from thebasal cells by a clear zone of approximatelythe same width, and scattered along the lengthof the basal cell are thickenings of the cellmembrane which resemble one-half of a des-mosome and hence are called hemidesmo-somes. Filaments have been described crossingthis clear zone, but resolution of these filamentsis difficult because of their size and lack ofstaining. On the connective tissue side of thebasement membrane the collagen fibres form alacework which is apparenty connected to theamorphous lamina densa by a series of an-choring fibrils' through which collagenousfibres run. The exact location of the areastaining with PAS in the light microscopehas not been identified at the ultrastructurallevel, though it would seem likely that thewhole complex-namely, the clear and densezones together with the anchoring filamentv-would reach the width required for visualiza-tion with the light microscope, especially aftercomplexing with the PAS reagents.The basement membrane region is obviously

important, as all material passing into or outof the epithelium must cross it. Very little,however, is known about its role in limitingthe passage of materials. The thickness of thebasal lamina-that is, the lamina densa-hasbeen reported to be inversely related to thedegree of keratinization, suggesting an inherentproperty of the epithelium to keratinize or not,but I will return to this later. In disease statesthe basal lamina has been reported to becomeduplicated and broken up and to show varia-tions in density and continuiity2, but a study

of such changes is difficult in that the plane ofsection in normal epithelium often producesapparent changes in its character.

Basal layerThis layer, known as the stratum germrinativ-um or stratum basale, has cells which undergodivision to make up for the cells shed off fromthe surface. The attachment of the cells to thebasement membrane by way of hemidesmo-somes has been mentioned. Thley are also incontact with their neighbours by desmosomesand between these there is a variable amountof amorphous material in the intercellularspace. Basal cells of human buccalepithelium show few differences comparedwith those of keratinizing epithelium. Theircontents of tonofilaments, their size, andtheir cellular organelles are similar inthe epithelia from the two areas. Wherethere does appear to be a difference is inthe morphology of the connective-tissue-epithelial junction. In some cases the interfaceconsists of a series of ridges and grooves ratherthan pegs as in the classical description. Theridges may be narrow, and this is more com-monly associated with keratinization, or broad,

FIG. 2 Electron micrograph of junction be-tween epithelium and connective tissue. Thelamina densa is separated from the epithelialcells by a clear zone. Hemidesmosomes arefound at intervals on the basal cell membrane.

Keratinization of the oral epithelium 353

as in non-keratinizing epithelium3. The effectof the ridges and pegs, when they occur, is toincrease the surface area either for attachmentor for exchange and on these grounds the inter-face is stronger in masticatory keratinizingmucosa than in the lining or non-keratinizingtype.The complexity of the interface has led to

confusion in relating mitotic activity in thebasal layer to the turnover of the epithelium.Studies of cellular proliferation have beenmade easier by the use of tritiated thymidine,a radioactive precursor incorporated exclusive-ly into the DNA of dividing cells. The daugh-ter cells retain radioactive DNA and thus it ispossible to follow daughter cells resulting froma cell division which occurred shortly afteradministration of this material. With this tech-nique attempts have been made to relate cell-ular proliferation rates to keratinization poten-tial in the mouths of animals and man. Gen-erally speaking, the non-keratinizing tissueshave a higher turnover rate than the keratiniz-ing epithelia, and the time taken for a dividingcell to differentiate and pass through to thesurface in the mouse palate is 6-7 days4. Onthe cheek the rate is 32-4 days and for com-parison it has been calculated that turnoverin skin is of the order of 30 days, though indiseased states such as psoriasis this may bespeeded up dramatically. Man has notbeen studied in any detail as yet. An interestingaspect of this problem is what determines whichcell is to be released from the pool of dividingcells. This is apparently not simply a case ofone daughter cell being pushed upward, butsome evidence suggests that both daughtercells may at times be retained and a restingcell pushed into the maturation pool5.

There is some doubt whether or not celldivision occturs above the basal layer in thenormal adult epithelium (Fig. 3). Mitotic fig-ures and tritium-labelled cells are certainlyfound there, but this could be due to the sec-tion plane passing tangentially through a pegof connective tissue. Suprabasal mitoses makeup a very small percentage of the whole divid-ing population in the mouth. In skin supra-basal mitoses have been estimated by Halprin'as making up one-third of the total number.The control of mitotic behaviour in the

basal layers inight be thouight to determine

the degree of keratinization, since it has beenseen that non-keratinizing areas have a higherturnover rate. A feedback mechanism has beendescribed in skin, in which it was shown thatstripping off the superficial layers with adhesivetape could dramatically increase basal pro-liferation rates some 40 times7. Bullough andLaurence8 discussed the mode of action ofchalones produced by the epithelium and dem-onstrated that in combination with adrenalinethey could inhibit mitosis. If increased loss ofepithelial cells occurs owing, for example, toabrasion, then an increase in mitotic activitywhen released from inhibition makes up forthis loss. This activity will eventually increasethe level of chalone production and thus themitotic activity will again be inhibited asnormal thickness is reached. The precisenature of the chemical process involved is notknown, nor has the effect of the chalones onkeratinization been investigated.

Stratum spinosumThe next layer is the stratum spinosum orprickle-cell layer. This layer, in contrast tothe basal layer, is many cells thick and makesup the bulk of the epithelium. The cells arecharacterized by the development of intercell-r-S~~~~~~~~ A* Ao..FIG. 3 Autoradiograph of mouse palate 4 hafter injection of tritiated thymidine. Labelledcells are found in the basal layer and sometimesappear in the layer above this. Haematoxylin.

354 David Adams

- .1 11I#S the differences are not striking and the situa-tion varies from one animal species to another.

Straturn granulosumThis layer is so called because of the granulesof keratohyalin it contains, and these can beseen with the light microscope in palatal ortongue epithelium. Chemical analysis of iso-lated keratohyaline9 indicates that this is a sul-phur-r-ich amorphous precursor of the hornycell content rather than a side product of thekeratinization process. The ultrastructuralcharacteristics of the keratohyaline granuleshave complicated rather than simplified thepicture in relation to their function. At leasttwo types of granule exist'0. One type, seenin highly keratinized areas, is irregular in

- sk .X. shape, highly electron dense, and associatedwith tonofilaments (Fig. 5). These are also seen

FIG. 4 Electron micrograph of stratum spino- in skin epithelium. This type is not found insum. Cells meet each other across the inter- non-keratinizmg areas of the human mouthcellular spaces at desmosomes. Insert-light- but may occur in parakeratinized areas of themicroscope section of this region for com- palate and gingiva. A second type is moreparison. Haematoxylin and eosin. regular in outline and has been described by

Jessen" in both keratinizing and non-keratin-izing epithelia. It does not appear to be asso-

ular space and by spinous processes or pricklesall around their periphery (Fig. 4). Electron itmicroscopy has revealed that the cell contentsare not confluent where the spinous processes

`4touch, but here the cells are attached to eachother by desmosomes.At these 'spot welds' there is very close con-

tact of the cell membranes, and the inner leafletof the cell membrane is thickened. Tonofila-ments abound in the cells and many run into AW.

the dense plaque of the desmosome. It maybe that they link up with other plaques, but _it is not clear if any filaments actually cross - Dthe gap between cells at the desmosome.Although the numbers of desmosomes aremuch greater in the stratum spinosum thanin the stratum basale, it is rare to find desmo-somes in the process of formation. If desmo-somes are attachment plates, then they must vbreak down and re-form as individual cellsmove past one another, as the autoradio- FIG. 5 Electron micrograph of stratum granu-graphic evidence would suggest takes place, losum in keratinizing epithelium with typicalbut no reports of such breakdown and re- keratohyaline granule associated with tono-formation in oral epithelium have been found. filaments. (Reproduced from Chen andTonofilaments tend to accumulate more in the Meyer"0 by courtesy of Charles C Thomas,keratinizing tissue, but in the stratum spinosum Publishers, Springfield, Illinois.)

Keratinization of the oral epithelium 355

FIG. 6 Electron micrograph of cell fromstratum granulosum showing the second typeof keratohyaline granule. Small granules areembedded in an electron-dense matrix.

ciated with tonofilaments (Fig. 6). Jessen, byusing oxidation and pepsin digestion, con-cluded that this second type was made upof two components, 'single' granules and adense matrix in which they were embedded.The 'single' granules he believed to contributeto the thickening of the cell membrane whichis seen in keratinizing cells, and the matrix wasthought to become dispersed between thetonofilaments. However, there does seem to beconsiderable species variation in the morpho-logy of the granules, and the function ofsimilar granules in non-keratinizing epitheliumis problematical.

Another type of granule is found in thecells of the upper layers of the stratum spino-sum and in increasing numbers in the cells ofthe stratum granulosum. These granules, vis-ible only with the electron microscope, measureabout O.I-0.2 ,um across and have been thesubject of many investigations. They werecalled 'membrane-coating granules'12 as it wasconsidered that they produce the thickening ofthe cell which is seen in keratinized epitheliaThey have also been named microgranules,keratinosomes, and lysosomes. They have acircular or elongated profile and contain eitheramorphous or stacked lamellated material

(Fig. 7). Martinez and Peters'3 believed thatthe lamellated granules contained discs with afive-layered structure and that after dischargeinto the intercellular space they were appliedto the outer cell surface. This modificationstrengthened the superficial surface of the cellsand perhaps contributed to the barrier func-tion of the epithelium. Electron histochemistryreveals acid phosphatase within the granule,and the granules are thought to shed theircontents into the intercellular spaces'4 sinceacid phosphatase activity is found both in thespace and in intercellular structures which re-semble the stacked membranes. The acid phos-phatase is considered by some workers to acton the desmosomes, thus facilitating the releaseof cells from the surface'5.The differing appearance between the

microgranules of keratinizing epithelia, wherethe lamellated type are found, and those ofnon-keratinizing epithelia, in which the inter-nal structure of the microgranule is usuallyamorphous, suggests different functions in thetwo situations. Hashimoto'6 believed that theamorphous granules contributed to the inter-cellular cementing substance. They have beendescribed in large numbers in developing oralmucosa, where they apparently contribute to

FIG. 7 Microgranules in the stratum granu-losum. They are spherical and sometimes havean amorphous content. Insert-lamellatedmicrogranule.

356 David Adams

the extracellular 'fuzz'. Another interpretationis that the granules are not separate vesiclesbut inpouchings of the cell membrane'7. Thecellular interdigitations are so complex that inplaces they are sectioned transversely and mayappear to be within the cell. Support for thissuggestion comes from the disposition of thegranules near the superficial periphery of thecell. It is further suggested that the granulecontents are similar to the contents of theintercellular space because they are part of it.The acid phosphatase appears to be inside themicrogranule only because it is present in theintercellular space.A very recent scanning electron microscope

study'8 of the surface characteristics of theepithelial cells indicates that there are pits inthe surface of the cells as well as microvilli,and this suggests that at least some of the ap-pearances of the microgranules may indeed bedue to the plane of section.

Stratum corneumIn this layer there is a dramatic change in thestructure of the cells. The nucleus disappears,the granules are lost as well as all the organ-elles, and the cell appears to be packed withtonofilaments. Changes are seen also in theintercellular space. Desmosomes lose their in-termediate layer and many are lost altogethersince the numbers are reduced. Often there issimply a dense zone between the cells. As indi-cated earlier, the cell wall thickens by anincrease in thickness of the inner cell laminaand the profile of the cell becomes much lesscomplicated as many of the microvilli are lost.The internal architecture of the cells is highlyelectron dense (Fig. 8), which has led todifficulties in investigating the keratinizingprocess. Packing of tonofilaments appears tovary randomly among the cells in a similarway to the variations reported in skin keratin'9.

Meyer'" has investigated the tonofilamentsin keratinizing and non-keratinizing epitheliaand concludes that the packing and orienta-tion of these fibrils within the cytoplasm arethe most distinctive differences between thetwo types. She points out, however, that therewould appear to be a spectrum ranging fromnon-keratinizing epithelia through parakeratin-ization to orthokeratinization.As the cells become condensed and flattened

a distinct thickened cell membrane appears.Thus keratinization is not a change that is con-fined to the cell contents but includes modifi-cation of the cell membrane and the intercell-ular material. This modification means thatalthough the cell membrane is physicallyweaker than keratin, chemical processes areresisted more than with keratin and thus thecomified layer could be regarded as an effi-cient, well-integrated system of protection tothe underlying tissues.The mystery of the disappearing nucleus

and other organelles at the beginning of thislayer has not been solved. No trace is foundin light or electron microscope studies. Per-haps the fixatives we use dissolve out the par-tially disintegrated and soluble products of theirbreakdown. There is a variation in the degreeof disintegration depending on the degree ofkeratinization, but Omanski"0 sums up thesituation by saying: 'Little is known about thebreakdown other than the fact that it occurs'.

Another interesting feature of the stratumcomeum is the passage of cells through it. Nolonger do we see the random migration ofindividual cells, but all cells apparently mi-grate as a sheet. The evidence for this comesfrom autoradiographic studies with labelledprotein. The radioactivity is found in a band

FIG. 8 Stratum corneum. The cells havethickened cell membranes and contain densebundles of tonofilaments.

Al- ..

Keratinization of the oral epithelium 357

which moves up through the stratum corneum

uniformly. At the surface, rather than indi-vidual cells being shed, sheets of the superficiallayer are lost, this loss being a reflection of thedegree of wear and tear on the surface.Changes in desmosome structure may be asso-

ciated with the loss, but there are conflictingreports on these structures and the releasemechanism has still to be explained.

During keratinization the water content ofthe cells decreases dramatically in spite of thefluid environment in the mouth. Thus it is notjust desiccation of the surface layers, as may

occur in skin. Work is done in expelling thewater from between the bundles of tonofila-ments, though our knowledge of this process isscanty. It has been estimated2' that in passing

from the basal layer to the surface of the gran-

ular layer a cell increases its dry weight 29-

fold. The observations were based on the buc-cal cells in the rat and this is similar to therate of protein synthesis in the pancreas.

Keratin in the mouth remains relativelytranslucent as compared with skin, probablybecause it is kept moist and because the corni-fied layers are normally compact. In skin thesuperficial layer of the stratum corneum, thestratum disjunctum, a layer of loosely knitbundles of cornified cells, gives the skin itsopacity. In the mouth if the layers of keratinbecome broken up or if abnormal keratinizationoccurs there is a loss of translucency and thearea appears white, perhaps because of thespaces in the cornified layer.

Regulation of keratinizationThe morphology of keratinization in the mouthhaving been dealt with, it is pertinent to lookat the factors that govern the process and thatmight lead us to methods of controlling or

altering this mysterious property of the epi-thelium. The degree of keratinization-thatis, the thickness of the keratin layer andwhether para- or orthokeratinization occur-

is dependent on several factors.The first of these is genetic. The epithelium

of the cheek and the palate show differencesas early as 15-I6 weeks in utero. Peridermcells which are associated with keratinizationof skin during development are found in theoral cavity only on those sites that are destinedto become keratinized. Further keratinization

occurs on the palate before birth, indicatingan independence of functional stimuli.

Attempts to alter the degree of keratiniza-tion have met with ambiguous results. Tooth-brushing and other methods of producing fric-tion will not induce the normally non-keratin-izing epithelium of either the alveolar mucosaor the lining mucosa of the gingival sulcus tokeratinize. On the other hand gingival mucosawhich shows parakeratinization will changetowards orthokeratinization with frictionalstimuli. It is interesting in this respect to notethat inflammation may produce a changefrom ortho- to parakeratinization, and tooth-brushing may simply remove the irritant thatis inducing the inflammation. With transplan-tation experiments it is difficult to know howlong the transplanted epithelium survives, butrecent elegant experiments have shown theimportance of the underlying connective tissue.Grafts from the palate stripped of all epi-thelium and implanted on the alveolar bonebecame covered by keratinized epithelium afterhealing, whereas similar connective tissue graftsfrom the buccal sulcus had non-keratinizingepithelium covering them on healing22. A simi-lar result has been achieeved with mouse skin.Spearman2" transplanted morphologicallydistinguishable mouse ear epidermis, aftercombining it with tail dermis, into a preparedbed in the tail of a second mouse. After 30days the transplanted epidermis was alteredfrom the ear type to the tail type.

It has also been suggested that systemic fact-ors affect the degree of keratinization. A recentsequential cytological study24 over 2 monthsshowed no differences between young menand women in the variation of the cornifica-tion of the palatal epithelium. Another studyon the keratinization of the gingivae duringpregnancy and afterwards has likewise shownthat any changes that occur are more likelyto be the result of local inflammatory condi-tions than of any systemic hormonal varia-tions25.

Dendritic cellsIn addition to the keratinoblasts in the basallayer and the keratinocytes in the functionallayers, dendritic cells of two types have beendistinguished. These are called melanocytesand Langerhans' cells. They resemble each

358 David Adams

other in structure but differ in cell content andin position. Melanocytes are found among thebasal cells and contain granules of melanin.They are present in all races and are probablyof neural crest origin. The Langerhans' cellsare found in more superficial layers and con-tain characteristic 'tennis-racket' organelles.These cells were thought to be effete melano-cytes, but recently it has been suggested thatthey may control keratinization, since they aremore numerous in keratinizing epithelia26. Anassociation with chalone production has alsobeen postulated.

ConclusionIn summary, then, the oral epithelium is seento show a wide variation in the degree towhich the surface cells become keratinized.The basement membrane and the connectivetissues below this seem to exert an influenceon this process of keratinization. Basal cellsproliferate more slowly in oral keratinizingregions but faster than in epidermal sites.During keratinization tonofilaments accumu-late and are arranged in dense bundles. Thebundles are embedded in an amorphous cementwhich may have its origin in the kerato-hyalin material. Modification of the cell mem-brane is an essential part of keratinization andthis occurs by thickening of the inner leafletand addition of material to the outer aspectof the cell. This latter material appears to bederived from the microgranules, which mayalso be concerned in loosening the cells beforedesquamation.The whole process appears to be genetically

determined and the underlying connectivetissue may play a role in this process in theadult. Local factors can alter to a limited ex-tent the degree of keratinization of the surfacecells. It can be seen, therefore, that keratiniza-tion is an intriguing, complicated, and import-ant process whose true nature still awaits tobe explained. This explanation, I believe, willcome only through the collation of informa-tion from biochemical, physiological, andmorphological studies.

ReferencesI Susi, F R (I969) Journal of Dental Research, 48,

'44.2 Takarada, H, Cattoni, M, Sugimoto, A, and

Rose, G G (I974) journal of Periodontology,45, 288.

3 Loe, H, and Karring, T (I971) ScandinavianJournal of Dental Research, 79, 315.

4 Hamilton, I A, and Blackwood, H J J (I974)Journal of Anatomy, II7, 3I3.

5 Pereira, J P M, and Leblond, C P (I965)American Journal of Anatomy, I17, 73.

6 Halprin, K M (1972) British Journal of Derma-tology, 86, 14.

7 Pinkus, H (I952) Journal of InvestigativeDermatology, I9, 431.

8 Bullough, W S, and Laurence, E B (I96I)Proceedings of the Royal Society, Series B, I54,540.

9 Matoltsy, A G, and Matoltsy, M N (I970)Journal of Cell Biology, 47, 593.

Io Chen, S Y, and Meyer, J (97I) in Current Con-cepts of the Histology of the Oral Mucosa, ed.Squier, C A, and Meyers, J, ch. 6. Springfield,Ill., Thomas.

ii Jessen, H (I970) journal of UltrastructureResearch, 33, 95.

I2 Matoltsy, A G, and Parakkal, P F (i965) Journalof Cell Biology, 24, 297.

13 Martinez, I R, and Peters, A (I970) AmericanJournal of Anatomy, I30, 93.

14 Weinstock, M, and Wilgram, G F (1970) Journalof Ultrastructure Research, 30, 262.

I5 Wilgram, G F (I966) Archives of Dermatology,94, I27.

i6 Hashimoto, K, and Lever, W F (I967) Journalof Investigative Dermatology, 48, 540.

17 El-Labban N G, and Kramer, I R H (1974)Journal of Dental Research, 53, 1071.

i8 Cleaton-Jones, P (I975) Journal of PeriodontalResearch, IO, 79.

1g Brody, I (I970) Journal of UltrastructureResearch, 30, 6oi.

20 Omanski, C P (I97I) in Current Concepts of theHistology of the Oral Mucosa, ed. Squier, C A,and Meyers, J, ch 7. Springfield, Ill., Thomas.

2I Meyer, J, Alvares, 0 F, and Barrington, E P(1970) Growth, 34, 57.

22 Karring T (1974) Personal communication.23 Spearman, R I C (1974) Acta Anatomica, 89,

195.

24 Lindholm, K (I975) Proceedings of the FinnishDental Society, 71, 84.

25 Hugoson, A, Winberg, E, and Angstrom, T (I971)Odontologisk Revy, 22, 145.

26 Sagebiel, R W, Clarke, M A, and Hutchens, L H(I971) in Current Concepts of the Histology ofthe Oral Mucosa, ed. Squier, C A, and Meyers, J,ch. 8. Springfield, Ill., Thomas.