0022-202X/84/8306-04 :l l $02 .00/ 0 TH E .] Q URNAL OF (NVESTIC.ATIVE DERMATOLOC:Y, 83: 4:.!1 - 4:35 , !984 Vol. 83. No. G Cop y right. (Q 1984 by The Williams & Wilkins Co. Printed in U.S. A. The of Rabbit Normal Epidermal Merkel Cells Are

JEAN-HILAIRE SAURAT, M.D., LILIANE DIDIERJEAN, PH.D., 0MAR SKALLI, PH.D., GEORGES SIEGENTHALER, PH.D., AND GUILIO GABBIANI, M.D. Departments of Dermatology (J -HS, LD, GS) and Pathology (OS, GG), Faculty of Medicine, Geneva, Switzerland

Four hundred Merkel cells (MC) have been studied by As MC are considered to on gmate from neural cells [1] it double-label immunofluorescence using: (1) a mono­ fo llows t hat t hey should express either NF or GFA proteins. clonal which has been previously demonstrated We have recently shown that MC express neither NF [4] nor to react with MC and (2) antisera and monoclonal anti­ GFA proteins [5] and we demonstrate here that, like epithelial bodies against the 5 types of intermediate filaments. It cells, normal rabbit MC express cytokeratins. was d emonstr ated that MC did not react with , , glial acidic fibrillary , or antisera. A strong staining of MC was observed with 2 MATERIALS AND METHODS antisera and 2 monoclonal against . Rabbit lip substrate, in which MC are readily detectable, has been The polypeptide pattern of MC is probably described in detail previously (2] as well as t he monoclonal antibody similar to that of simple epithelia. These findings attest reacting with MC (MCab) . Seri al cryostat sections (4 1.11n) of fro ze n to the epithelial nature of MC. rabbit lip specimens were prepared from 10 rabbits and grouped in series of 3 successive sections. In each series, the 3 s uccessive sections were processed as fo llows.

M erkel cells (MC) are thought to be specialized mechanore­ Localization of M C with MCab (One-Step Indirect ceptor cells present in the skin [1] ; as the population of MC is I m.munofluorescence) very small and there exists no s pecific staining method for The first section was incubated with the M t hem, the only way of identifying MC is electron Cab diluted 1:100 and microscopi­ t hen a rhodamine-labeled goat antihuman kappa light c hain (Behring; cally. This drawback might e xplain the limited knowledge batch no. A 125003A working dilution 1:80). available concerning t his cell, whose origin and function still remain unknown more than a century after its original descrip­ S tudy of the Staining Patterns wit.h Various Antisera and Monoclonal t ion by Merkel. Antibodies Against IF Proteins (One-Step Imm.un.ofluorescen.ce) We recently described a monoclonal antibody reacting wit h The second section was in cubated with an antiserum to one of the rabbit MC [2]; this offered a new tool for the study of several fiv e IF proteins: aspects of MC biology. A feature of MC which could help to l. Keratin: Two antisera and three monoclonal antibodies were used: unde rstand their origin would be to elucidate the nature of the Antisera KG I (affinity purified antiserum against heratin) and KG2 protein(s) constituting the 10-nm intermediate-sized filaments (antiserum against keratin): KG1 was raised in guinea pigs injected (IF) v isible wi thin the cell by electron microscopy [1]. This with bov ine hoof prekeratin. Affinity purified a ntibodies were prepared appeared of crucial importance because IF of different cells, by passing t he serum through a column of glutaraldehyde-activated ultrogel AcA22 coval a lt hough similar in morphology, show differences in protein ently linked [6] with pure bovine hoof prekeratin [7]; KG1 was used at a concentration of 60 11g/ ml a nd revealed with a composition in cells having different embryologic origins [3]. flu orescein -labeled goat anti-guinea pig immunoglobulin antiserum Five different types of IF are c urrently recognized in verte­ (Cappel; 1:20). KG2 was not affinity purified; it was raised in !!llinea brates: cytokeratins in epithelial cells, vimentin in cells derived pigs injected with bov ine hoof prekeratin, used at a dilution of 1:40 and from embryonal mesenchyme, desmin in muscle cells, neuro­ revealed as KG l. filaments (NF) in neuronal cells and adul t neural tissues, and Monoclorwl antibodies: KLJ is a mouse monoclonal antibody obtained finally glial fibrillary acidic (GFA) protein characteristic of by immunization with human keratin [8]. It has been show n to react. astrocytes and Bergmann glia. This distribution is striking by immunoblotting with keratin polypeptides of 67, 63, 57, and 55 because it follows well-known histologic principles. Thus, ktlodaltons (kD) e xtracted from huma n epidermis [8]. KL1 was used at a dilution of 1:10 and revealed wi th a flu oresceinated goat antimouse knowledge of the type of IF expressed in a cell can allow the immunoglobulins antiserum (Sigma; 1:150). assessment of its origin [3]. RGE53 is a mouse monoclonal antibody obtained from BALB/ c mice immunized with H eLa cell ; it has been shown to react in immunoblotting with a cytoskeleton associated polypeptide with a Manuscript received March 20, 1984; accepted for publication June molecul ar weight of approx imately 45,000 in both HeLa cells a nd 13, 1984. cul tured human h epatoma cells [9]. RGE53 was used undiluted and Supported in part by g rants from t he Fonds National Suisse de revealed as KLl. Recherche Scientifique No. 3957082 and No. 3178082. PKI( J is a monoclonal antibody raised against keratin isolated from Reprin t requests to: J . H . Saurat, Department of Dermatology, a pig kidney epit heli al cell line [10]. It has been show n to react in Hopital Canto nal Universitaire, 1211 Geneve 4, Switzerland. immunobl otting with keratin polypeptides of 41, 45, 48, and 56 kD Abbreviations: extracted from a pi g kidney epit helia l ce ll line [1 0]. PKK1 was used at G FA: gli al fibrill ary acidic (protein ) a dilution of 1:10 and revealed as KL1 and RGE53. IF: in termedi ate-sized fil aments These t hree monoclonal antibodies against keratin did not react with kD: kilodaltons other IF proteins [8- 10]. KG 1: affinity purified antiserum against keratin 2. Neurofila.m.ents: A rabbit polyvalent neurofilament antiserum (dil KG2: antiserum against : 20) reacting with the 3 polypeptides of 200, 150, and 70 kD [11] was KL1: monoclonal antibody against keratin used as previously described (4] . It was revealed wit h a flu orescein ­ M C: Merke l cell (s) labeled g-oa t anti rabbit immunoglobulin serum (Behring; 1:20) . M Cab: Merkel cell monoc lonal ant ibody 3. Vim entin: A monospecific gu inea pig a ntiserum to viment.in [1 2] NF: neurofi lament was used at a concentration of 50 pg/ ml and revealed with a fluorescein­ Pl

431 432 SAURAT ET AL Vol. 83, No.6

Ftc. l. Cryostat sections of rabbit lip. a and b, One·step indirect immunoflu orescence wi th the monoclonal antibody RGE53, showing brightly /1 ua re~ce n t cell s (arrow) in the region where MC usually c lu ster (see [2]) (a x 100; b x 400). Note that RGE53 does not react with other cells within either epiderm is o r dermis. c and d, Double-label immunofluorescence (X 1000). c shows the reaction of RGE53 antibody revealed with a fl uorescein-labeled antiserum. d shows t he reaction of MCab revealed with a rhodamine-labeled antiserum. Most MC appear to be RGE53- positive but two of them are not stained b y RGE53 (see Discussion). e and[, Dou ble-label immunoflu orescence (X 250). e shows t he reaction of KLl antibody revealed with a flu orescein -labeled antiserum. d shows the reaction of MCab revealed with a rhodamine- labeled antiserum. Note that MC are cl ea rly KLl -negative (arrowheads to locali ze 3 MC) . at a co nce ntration of 33 pg/ml and revealed with a flu orescein-labeled mi xed at a fin al dilu tion identical to that of t he simple staining and goat antirabbit immunog lobulin serum (Behring; 1: 20). in cubated wit h the specimens for 45 min. After washing for 15 min, a !'>. Glial acidic fibrillar protein.: A rabbit serum against GFA protein mixture o f the relevant rhodamine- and flu orescein -labeled antisera (OAKO Co rporal ion , Santa Barbara, California; 1: 10) was used and used in the simple staining was applied to t he spec imens for 20 min. revealed with flu orescein -labeled swin e a ntirabbit immunoglobulin The s lides were then mounted after a fin al washing procedure of 15 se rum (Behrin g; I :20). min. All t he dilutions were done in phosphate-buffered saline, pH 7.4, with 0.2 % bov ine serum albumin and 0.03 % Triton-X-100. The entire Double- Label fm.nwnofluorescen.ce immunolluorescence procedure was conducted according to standard The third section was processed for double stainin g: both MCab and techniques and proper co nt rols were included as previously desc ribed one of the a ntisera or monoclonal antibodies against IF proteins were 12]. Dec. 1984 CYTOKERATINS IN MERKEL CELLS 433

FIG 2. Double- label immunoflu oresce nce. a, c, and e show the reaction of MCab with MC (arrows) on the parakeratotic side of the rabbit. lip specimen revealed by a rhodami ne- labeled antiserum. b (x 400) illustrates the rea ct i ~n of the same area as A with the antivimentin serum revealed by a fluorescein -labeled antiserum; the ep1dermal MC (arrowheads) are not stamed whereas derm al !ibroblasts are. d (X 400) shows the reaction of the same area as c with the a nti-GFA serum revealed by a fluorescein -labeled antiserum; the epidermal MC (arrowhead) are not stained whereas a reaction is observed with dermal nerve processes. f (X 100) shows the reaction of the same area as e with the antidesm in serum revealed by a fluoresce in -labeled antiserum; the epidermal MC (arrowhead) are not. stained whereas dermal muscles are reactive.

RESULTS Although KG 1 and KG2 uniformly stained the entire epider­ mis, the presumed MC often appeared brighter than the sur­ One-Step Immunofluorescence rounding keratinocytes, especially in the parakeratotic side of the rabbit lip specimen. PKK1 also reacted with some kerat i­ Two antisera, KG 1 and KG2, and 2 of the 3 monoclonals, nocytes in the upper level of the parakeratotic . RGE53 and PKK1, directed against cytokeratins were found to RGE53 reacted only with these int raepidermal scattered cells react by one-step immunofluorescence with scattered intraepi­ with no staining of the surrounding keratinocytes (Fig 1a,b )_ dermal cells having the characteristic shape and distribution of KG1, KG 2, RGE53, and PKK1 did not react with nerve proc­ MC within the rabbit lip epithelium (Fig 1) . esses, muscles, or dermal fibroblasts. 434 SAURAT ET AL Vol. 83, No. 6

NF antiserum reacted clearly with dermal nerve processes as Such findings demonstrate that MC express a cellular com­ previously reported [4], but not with intraepidermal cells. Sim­ ponent characteristic of an epithelial cell. That MC is an ilarly, GFA antiserum stained only dermal nerve processes. epithelial cell might seem discrepant with the fact that MC Desmin antiserum reacted only with dermal muscles. Vimentin were s hown to contain neuron-specific enolase [13], Met-en­ antiserum reacted brightly with dermal fibroblasts; in the rabbit kephalin (14), and vasoactive-intestinal polypeptide [15) which lip only very few intraepidermal vimentin-positive cells, pre­ are characteristic of "neuroendocrine" cells. However, this sumably Langerhans cells and melanocytes, could be detected. would not be the first example of a cell expressing both cyto­ None of these cells had the shape and distribution of those keratin and biochemical properties of a neuroendocrine cell; in stained with RGE53 and PKKl. fact, cytokeratin was recently demonstrated in insulin-secreting islet cells [16]. MC are considered as immigrant cells that enter Double-Label Immunofluorescence the epidermis during fetal life although this has never been Four hundred MC, as defined by a bright reaction with MCab, clearly demonstrated [lJ. None of the other known intraepi­ were analyzed. None was found to react with either NF (see dermal immigrant cells expresses cytokeratin; melanocytes, [4]) GFA, desmin, or vimentin antisera (Fig 2). Similarly KLl Langerhans cells [17,18], and the Thy-1 + cells in the mouse did not react with MC (Fig 1). [19) express vimentin. In contrast KGl, KG2, RGE53, and PKKl clearly reacted The suggestion t hat t he MC is a specialized or modified with MC. (Fig 1). It was found t hat about 98% of MCab­ keratinocyte may no longer "go against common sense" [20]. positive cells were KGl-, KG2-, and PKK1-positive as well. On the contrary, since the epidermal MC is epithelial in nature, About 95% of MCab-positive cells were found to be RGE53- it might go against common sense to consider it as an intra­ positive. Careful analysis of the slides (see Fig lb,c) as well as epidermal immigrant cell. This would possibly reorient the t he results of the one-step immunofluorescence clearly estab­ concepts concerning this cell and aid in establishing its func­ lished that the staining of MC by these keratin antibodies was tion. not due to background fluorescence of MCab. The lack of We thank D. staining of MC in the double-label immunofluorescence with Dahl for providing NF antiserum. The illustrations were done by M. Pisteur and the manuscript typed by C. W asung and KLl (see Fig l e, f), as well as vimentin, desmin, and GFA C. Fonjallaz. antisera (see Fig 2) further supported this fact. REFERENCES DISCUSSION 1. Breathnach AS: The mamm alia and avian Merkel cell, The Skin Our results demonstrate that the intermediate filament pro­ of Vertebrates. Edited by RIC Spearman, PA Riley. Linnean teins o f normal intraepidermal MC are cytokeratins in the Society Series 9. Dorchester/Dorset, Henry Ling Ltd, 1979, pp rabbit lip. 283-291 2. Saurat JH, Chavaz P, Carraux P, Didierjean L: A human mono­ Firstly, KG1, KG2, PKK1, and RGE53 reacted, in one-step clonal antibody reacting with Merkel ce lls: immuno!l uorescence, immunofluorescence, with intraepidermal cells having the very immunoperoxidase, and immunoelectron microscopy. J Invest characteristic shape and distribution of MC within the rabbit Dermatol 81:249-253, 1983 lip epithelium (see [2] for electron microscopy of these cells). 3. Osborn M, Geisler N, Shaw G, Sharp G, Weber I<: Intermediate filaments. Cold Spring Harbor Symp Quant Bioi 46:413-430, Secondly, the double-label immunofluorescence demonstra­ 1981 ted that these keratin-positive cells were also stained by the 4. Saurat JH, Merot Y, Didierjean L, Dahl D: Normal rabbit Merkel MCab which has been previously demonstrated by immunoe­ ce lls do not express neurofilam ent proteins. J Invest Dermatol lectron microscopy to react only with MC in these areas 82:641-642, 1984 of the 5. Saurat JH, Merot Y, Didierjean L, Dahl D: Normal rabbit Merkel rabbit lip [2). That 2-5% of MCab-positive cells were not ce ll s (MC) do not express the intermediate fil ament proteins of stained by the antisera or monoclonal antibodies against kera­ glial or neuronal ce lls (abstr). J Invest Dermatol 82:424 1984 tin might either be due to technical problems or illustrate a 6. Nag! RB, McDaniel KM , Clark VA, Payne CM: The use' of anti­ heterogenity in the keratin keratin antibodies in the diagnosis of human neoplasms. Am J pattern among the MC population. Clin Pathol 79:458-466, 1983 We could not yet establish, for example, whether a RGE53- 7. Franke WW, Weber K, Osborn M, Schmid E, Freudenstein C: negative MC is e ither KGl- or PKK1-positive. Antibody to prekeratin: decoration of tonofilament-Iike arrays Thirdly, MC were found to express none of the 4 other IF in various ce lls of epithelial character. Exp Cell Res 116:429- proteins. This observation strongly supports the finding of 445, 1978 8. Viae J, Rea no A, Brochier J, Staquet MJ, Thivolet J: Reactivity reactivity with antikeratin antibodies because it is a well­ patte rn of a monoclonal antikeratin antibody (KL1). J Invest established fa ct that nonneoplastic cells do not express more Dermatol 81:351-354, 1983 than one type of IF protein in vivo [3). 9. Ramaekers F, Huysmans A, Moesker 0 , Kant A, Jap P, Herman The precise analysis of the cytokeratin polypeptide pattern C, Vooijs P: Monoclonal antibo~y to keratin filam ents, specific for glandular ep1theha and the1r tumours. Use in surgical pa­ of MC will become possible when the isolation of MC reaches thology. Lab Invest 49:353-361, 1983 a high degree of purity. However our present data with a panel 10. Holthiifer H, Miettinen A , Paasivuo R, Lehto VP, Linder E, of monoclonal antibodies provide some information on this AIJ'than 0 , Virtanen I: Ce llular origin and differentiation of renal issue. MC were KL1-negative but PKKl- and RGE53-positive; . A fluoresce nce microscopic study with kidney·spe­ ci!ic antibodies, anti-inte rmediate !ilament antibodies and lee - this would suggest t hat MC do not express polypeptides of 67, tins. Lab Invest 49: 317-326, 1983 ' 63, 57, or 55 kD revealed by KL1 [8) but would express some 11. Dahl D, Selkoe DJ, Pero RT, Bignami AJ: Immunostaining of polypeptides in the range of 41 , 45, 48, and 56 kD revealed by neurofibrillary tangles in Alzheimer's se nile dimentia with a PKKl [10], and of 45 kD revealed by RGE53 [9) . neurofilament antiserum. J N eurosci 2:113-119, 1982 This is, 12. Franke WW, Schiller DL, Moll R, Winte r S, Schmid E Engel­ however, an approximation because t he reactivity of these brecht I, Denk H, Krepler R, Platzer B: Diversity of cytokeratins. monoclonal antibodies has been analyzed on extracted Differentiation specific e xpression of cytokeratin polypeptides from human epidermis [8) and cultured epithelial cells (9,10). in epithelial cells and tissues. J Mol Bioi 153:933- 959, 1981 It cannot be excluded that they would not react with polypep­ 13. Gu J , Polak JM, Tapia FJ, Marangos PJ, Pea rse AGE: Neuron specific enolase in the Merkel ce ll s of mammalian skin . Am J tides of exactly the same molecular weights if the cytoskeletton Pathol 104:63-68, 1981 proteins of MC were used as substrate. At any rate it is clear 14. Hartsc huh W, Weihe E, Buch ler M, Helmstaedter V, Feurle GE, t hat MC express some keratin polypeptides which are not Forssmann WG: Met-enkephalin lik e immunoreactivity in Mer­ expressed in t he keratinocytes. Our observations strongly sug­ kel cells. Cell Tissue Res 201:343- 348, 1979 15. Hartschuh W, Weihe E, Yanaihara N, Reinecke M: Immunohis­ gest that t he cytokeratin pattern of MC is s imilar to that of tochemical localization of vasoactive intestinal polypeptide simple epithelia [1 2). (VIP) in Merkel cells of various mammals: evidence for a neu- Dec. 1984 CYTOKERATINS IN MERKEL CELLS 435 romodulator function of the Merkel cell. J Invest Dermatol 398:119-128, 1982 81:361-364, 1983 18. Mahrle G, Bolling R, Osborn M, Weber K:Intermediate filaments 16. Schubart UK, Fields KL: Identification of a calcium-regulated of the vimenti n and prekeratin type in human epidermis. J Invest insulinoma cell phosphoprotein as an islet cell keratin. J Cell Dermatol81:46-48, 1983 Bioi 98:1001-1009, 1984 19. Tschachler E, Schuler G, Hutterer J , Leibl H, Wo lff K, Sting! G: 17. Loning T, Caseli tz J, Seifert G, Weber K, Osborn M: Identification Expresion of Thy-1 antigen by murine epidermal cells. J Invest of Langerhans cell s: simultaneous use of sera to intermediate Dermatol 81:282-285, 1983 filaments, T6 and HLA-DR antigens on oral mucosa, human 20. Breathnach AS: Branched cells in the epidermis: an overview. J epidermis and their tumors. Virchows Arch [Pathol Anat] Invest Dermatol 75:6-11, 1980