Friedrich Sigmund Merkel and his “Merkel cell”, morphology, development, and physiology: Review and new results



Merkel nerve endings are mechanoreceptors in the mammalian skin. They consist of large, pale cells with lobulated nuclei forming synapse-like contacts with enlarged terminal endings of myelinated nerve fibers. They were first described by F.S. Merkel in 1875. They are found in the skin and in those parts of the mucosa derived from the ectoderm. In mammals (apart from man), the largest accumulation of Merkel nerve endings is found in whiskers. In all vertebrates, Merkel nerve endings are located in the basal layer of the epidermis, apart from birds, where they are located in the dermis. Cytoskeletal filaments consisting of cytokeratins and osmiophilic granules containing a variety of neuropeptides are found in Merkel cells. In anseriform birds, groups of cells resembling Merkel cells, with discoid nerve terminals between cells, form Grandry corpuscles. There has been controversy over the origin of Merkel cells. Results from chick/quail chimeras show that, in birds, Merkel cells are a subpopulation of cells derived from the neural crest, which thus excludes their development from the epidermis. Most recently, also in mammals, conclusive evidence for a neural crest origin of Merkel cells has been obtained. Merkel cells and nerve terminals form mechanoreceptors. Calcium ions enter Merkel cells in response to mechanical stimuli, a process which triggers the release of calcium from intracellular stores resulting in exocytosis of neurotransmitter or neuromodulator. Recent results suggest that there may be glutamatergic transmission between Merkel cell and nerve terminal, which appears to be essential for the characteristic slowly adapting response of these receptors during maintained mechanical stimuli. Thus, we are convinced that Merkel cells with associated nerve terminals function as mechanoreceptor cells. Cells in the skin with a similar appearance as Merkel cells, but without contact to nerve terminals, are probably part of a diffuse neuroendocrine system and do not function as mechanoreceptors. Probably these cells, rather than those acting as mechanoreceptors, are the origin of a highly malignant skin cancer called Merkel cell carcinoma. Anat Rec Part A 271A:225–239, 2003. © 2003 Wiley-Liss, Inc.

Merkel cells are large clear oval cells in the skin of vertebrates. The vast majority of Merkel cells are intimately associated with a nerve terminal. They were first described by Friedrich Sigmund Merkel in 1875 and were referred to as “Tastzellen” or “touch cells.” Clusters of Merkel nerve endings in glabrous skin were called “touch corpuscles” (Tastscheiben; Merkel, 1875). These terms indicate Merkel's assumption of a mechanoreceptor function of these structures. Around the turn of the century, these receptors were generally termed “Merkel'sche Tastzellen” and “Merkel'sche Tastkörperchen” to indicate their function as mechanoreceptors (Szymonowicz, 1895). Later the prefix “Tast-” was frequently omitted, and the cells were simply called Merkel cells. More recently, this term has also been used for cells of similar appearance but without contact to nerve terminals (Toker, 1972; Moll et al., 1984). Nowadays, most studies focus on suggested neuroendocrine functions of Merkel cells and their possible malignant transformation into Merkel cell carcinomas. This finding led us to re-emphasize the mechanoreceptor function of Merkel cells and their associated nerve terminals. After a long-standing controversy about the role of Merkel cells within the mechanotransduction process (Gottschaldt and Vahle-Hinz, 1981; Ikeda et al., 1994; Mills and Diamond, 1995; Senok et al., 1996), recent experimental evidence points to the direct involvement of Merkel cells in the transformation of mechanical stimuli to action potentials in the afferent nerve fiber (Chan et al., 1996; Senok and Baumann, 1997; Tazaki and Suzuki, 1998; Fagan and Cahusac, 2001). This review is based on our published and unpublished data on the ultrastructure of Merkel nerve endings, compared with the original drawings by Merkel, as well as developmental studies and physiological experiments carried out in our laboratories.


Merkel cells can be distinguished from other “clear cells” of the epidermis (melanocytes and Langerhans cells) by electron microscopy (Fig. 1). Characteristic features of Merkel cells are their oval shape, measuring 10–15 μm in the long axis, arranged mainly in clusters at the base of the epidermis in close contact with nerve terminals. Merkel cells bear spike-like protrusions, which interdigitate with the surrounding keratinocytes and are attached by desmosomes. They possess large lobulated nuclei. The cytoplasm contains, apart from common organelles, also neurofilaments, intermediate cytokeratin filaments, and many large (50–110 nm) dense-core vesicles accumulated near the junction with the nerve fiber (Iggo and Muir, 1969; Breathnach and Robins, 1970). The cytoplasmic membrane is closely apposed to the membrane of an axonal terminal, with areas of synaptic membrane specialization (Chen et al., 1973; Iggo and Findlater, 1984). The nerve terminal is packed with mitochondria and optically clear vesicles (Munger, 1965).

Figure 1.

Ultrathin section from the planum nasale of a domestic cat, showing a typical Merkel nerve ending consisting of a Merkel cell (M) with characteristic finger-like cytoplasmic processes (asterisk) and a discoid nerve terminal (T). The terminal contains a large number of mitochondria. Synapse-like contacts between Merkel cell and nerve terminal are marked by arrows.


Merkel cells contain the low molecular weight cytokeratins CK 8, CK 18, CK 19, and CK 20, but not the types characteristic of fully differentiated keratinocytes. They are used to identify Merkel cells in light microscopy (Saurat et al., 1984; Moll et al., 1984, 1995, 1996). Positive immunoreactivity for CK 8, 18, and 19 is also found in outer root sheath keratinocytes and some early fetal epidermal cells and, thus, is not specific for Merkel cells in general. Cytokeratin 20, however, is of particular value, because it is highly specific for Merkel cells in normal squamous epithelia of humans, pigs, and mice (Moll et al., 1995). However, in taste buds and various epithelia of the gastrointestinal tract, CK 20-positive cells are also found (Moll et al., 1995; Witt and Kasper, 1999).

General markers of neuroendocrine cells, such as neuron-specific enolase, protein gene product 9.5, synaptophysin, and chromogranin A, are found immunohistochemically in Merkel cells (Zaccone, 1986; Gauweiler et al., 1988; Hartschuh and Weihe, 1989; Dalsgaard et al., 1989; Ramieri et al., 1992).

In addition, dense-core granules of Merkel cells show immunoreactivity for a variety of substances, including vasoactive intestinal polypeptide, met-enkephalin, serotonin, substance P, pancreastatin, calcitonin gene-related peptide (CGRP), somatostatin and bombesin (Hartschuh et al., 1979, 1983; Gu et al., 1981; Alvarez et al., 1988; Garcia-Caballero et al., 1989a,; Cheng Chew and Leung, 1992; for detailed review, see English et al., 1992). These substances were prime candidates as neurotransmitters or neuromodulators at the postulated Merkel cell–nerve terminal synapse. However, to date, there is no evidence to support a transmitter role for any of these substances (e.g., Gottschaldt and Vahle-Hinz, 1982).


Merkel nerve endings are found in the skin and some parts of the mucosa of all vertebrates, including fishes (Fox et al., 1980), amphibians (Tweedle, 1978), reptiles (Crowe and Whitear, 1978), birds (Saxod, 1978b), and mammals (Munger, 1965). Their location is usually in the basal layer of the epidermis (Figs. 1–3), as well as the epithelium of mucosa originating from ectoderm (Merkel, 1875; Whitear, 1974; Lane and Whitear, 1977; Halata and Baumann, 1999). However, in birds, they are observed in the superficial layer of the dermis (Saxod, 1978a). Occasionally, dermal Merkel cells are also found in mammals, especially in man (Mahrle and Orfanos, 1974; Halata, 1979; Moll et al., 1986a). In some anseriform birds, Grandry corpuscles are found composed of cells resembling Merkel cells (Saxod, 1978a).

Figure 2.

Ultrathin section from a rete peg in the pig's snout. Several Merkel (M) cells with nerve terminals below are seen above the basal layer of the epidermis. Below the rete peg, within the dermis, two small Pacinian corpuscles (P) are visible. Their long axis runs parallel to the surface of the rete peg.

Figure 3.

Original drawings from Merkel's publication from 1875, illustrating “touch corpuscles” in different mammals. Each touch corpuscle consists of Merkel cells and nerve terminals originating from myelinated nerve fibers.


Mammalian Merkel nerve endings are found in the epidermis of hairy and glabrous skin. In primate glabrous skin, they are located in the sweat gland ridges in the basal layer of the epidermis, whereas in other mammals they are in the rete pegs. Especially well developed Merkel nerve endings are found in the nose of moles, and are called Eimer organs (Eimer, 1871). In hairy skin, they are found in hair follicles and in thickened parts of the epidermis between hair follicles, the so-called “Haarscheiben” (Pinkus, 1905) (see Figs. 3 to 5).

Figure 4.

“Touch dome” (T) from the eyelid of a rhesus monkey in semithin section. Merkel cells (arrows) are visible in the basal layer of the epidermis in the rete peg.

Figure 5.

Ultrathin section of a Merkel cell (M) in the epidermis of a touch dome from the human eye lid in contact with the basal layer of keratinocytes (K) of the epidermis. The nerve terminals (T) surround the Merkel cell like an egg cup.

Merkel Nerve Endings in Glabrous Skin

Merkel cells in the mammalian glabrous skin are always found in the basal layer of the epidermis (Figs. 1, 2). An original drawing from Merkel's study (1875) is shown in Figure 3. There are two types of glabrous skin in mammals. The first type (pegged skin) has solid epidermal pegs of different size anchoring the epidermis into the dermis. The dermis contains blood vessels separating the epithelial pegs. A typical example is found in the pig snout and planum nasale of mole and cat (Figs. 1, 2). In the basal layer of the epidermis, clusters of Merkel cells are found at the base of these pegs. Depending on the size of the peg, the number of Merkel cells in each cluster varies between 4 and 40 cells. All Merkel cells are in synaptic contact with discoid terminals of myelinated axons (3–5 μm in diameter), which lose their myelin sheath on entering the epidermis.

The second type of glabrous skin (ridged skin) is typically found on the tips of fingers and toes of primates and marsupials. Clusters of up to 10 Merkel nerve endings are found at the base of the epidermal ridges near the penetration of the sweat gland ducts (Halata, 1975). Oval Merkel cells are also seen in this location. Their long axis runs parallel to the skin surface, and the nerve terminals are oriented below the Merkel cells toward the basal lamina.

Merkel Nerve Endings in Hairy Skin

In hair follicles of velus hairs and guard hairs, Merkel nerve endings are found at the thickening of the follicle (the so-called bulge region) below the sebaceous gland. This part of the hair follicle does not change its form during the hair cycle, and the bulge region is also a stem cell niche of the epidermis (Oshima et al., 2001).

Between tylotrich hair follicles in the hairy skin (of some mammals), the epidermis forms epithelial pegs of different size and density, often referred to as touch domes (Straile, 1960; Fig. 4). These are only rarely found in man and basically are similar to the epithelial pegs in glabrous skin (see above), containing variable numbers of Merkel nerve endings (Figs. 4, 5).

Merkel Nerve Endings in Sinus Hairs

Clusters of specialized large hairs (vibrissae, whiskers, or sinus hairs) are found in all mammals (apart from man), mainly in the face, and innervated by the trigeminal nerve. They consist of a large hair follicle with strong sturdy hair embedded in a blood sinus. Fibers of mimic muscles are inserted into the capsule of the blood sinus. Figure 6a (taken from Merkel's publication, 1875) shows original drawings of cross- and longitudinal sections through a sinus hair. The hair follicles are richly innervated, and the nerve fibers end in a variety of different mechanoreceptors, especially in Merkel nerve endings (up to 2,000 in one follicle) and free nerve endings (Gottschaldt et al., 1973; Halata, 1975). In rodents, the sinus hairs are most important for spatial orientation and tactile scanning of their immediate environment, functions requiring an extensive neural representation by way of somatotopically organized barrels in the somatosensory cortex (Woolsey and van der Loos, 1970).

Figure 6.

a: Original drawings from Merkel's publication from 1875, illustrating a pig sinus hair in cross- and longitudinal section, with a row of Merkel cells supplied by a myelinated nerve fiber. b: Semithin longitudinal section through a rhesus monkey sinus hair from the upper lip, in low magnification, showing a sebaceous gland (G), the cavernous blood sinus (S), the connective tissue capsule of the sinus (C), and the hair bulb (B). The rectangle indicates the area shown enlarged in c. c: Ultrathin longitudinal section through the thickened portion of the sinus hair follicle below the sebaceous gland. Merkel cells (M) and nerve terminals (T) are arranged obliquely to the glassy membrane (asterisk) like the scales of a pine cone, with the Merkel cells always directed toward the glassy membrane. A lanceolate nerve terminal (L) in close contact with the glassy membrane is seen on the outer side.

Merkel cells are located in the basal layer of the epithelium of the hair follicle, in the thickened part below the sebaceous gland (Fig. 6b). This area is surrounded by the ring sinus. Merkel cells are arranged like scales in a pine cone, oblique to the basal lamina (glassy membrane), sending cytoplasmic processes of approximately 3 μm through the glassy membrane. All of them are in contact with discoid nerve terminals on the opposite side (Fig. 6c). The part of Merkel cells facing nerve terminals contains the typical dense-core (osmiophilic) granules. The myelinated axons (diameter approximately 5 μm) lose their myelin sheath and Schwann cells on penetrating the basal lamina. After branching several times, one axon can supply up to 50 Merkel cells. An intriguing pattern of innervation of Merkel cells in rat and cat whiskers recently has been described in detail by Ebara et al. (2002). Other Merkel nerve endings are found in the epidermis close to the penetration of the hair shaft in the rete pegs of the epidermis, similar to the description above.


As mentioned above, in mammals, the vast majority of Merkel cells are located in the epidermis just above the basal lamina. However, especially during development, dermal Merkel cells (Fig. 7) are occasionally found in the papillary layer of the mammalian dermis (Halata, 1981; Moll et al., 1986a; Narisawa et al., 1992). The Merkel nerve terminal complexes are covered by thin cytoplasmic lamellae of the terminal Schwann cell (after the axon has lost its myelin sheath; Fig. 8a,b). Also in the dermis, these cells have typical cytoplasmic processes extending between the glial capsule cells and are associated with nerve terminals. In hairy skin, they are found close to hair follicles and between hair follicles in the deeper dermis (Mahrle and Orfanos, 1974; Halata, 1990). In biopsies from patients suffering from scleroderma or alopecia areata (and totalis), increased numbers of Merkel cells are found in the deeper dermis in many regions of the body (Halata et al., 1977; Halata, 1990). On very rare occasions, dermal Merkel cells are found without contact to nerve terminals (Halata and Munger, 1981). However, the possibility of finding nerve terminals in serial sections cannot be ruled out.

Figure 7.

Ultrathin sections from the skin of the planum nasale (a,b) and sinus hair (c) of a 3-day-old cat. a,b: Two pictures from a series of 50 showing dermal Merkel cells (M) in contact with nerve terminals (T) wrapped by a thin layer (asterisk) of the terminal glial cell. a: A finger-like cytoplasmic protrusion (arrow) penetrates the basal lamina and anchors between keratinocytes (K) of the basal layer of the epidermis in a. b: Ten sections below, a cytoplasmic protrusion (arrow) contacts the connective tissue of the dermis. A part of the Merkel cell not covered by the terminal glial cell is in contact with a long thin protrusion from keratinocytes (so called “Wurzelfuesschen”), forming desmosomes (arrowhead). c: Merkel cell (M) in the dermis of a sinus hair, sending a protoplasmic protrusion (arrow) through the glassy membrane (G). Desmosomal contact between Merkel cell and keratinocyte (K) of the hair follicle is indicated by the arrowhead. In the neighbourhood, a typical lanceolate nerve ending with nerve terminal (L) covered by a lamella from Schwann cell anchored in the glassy membrane. Scale bar = 5 μm in b (applies to a–c).

Figure 8.

a,b: Ultrathin sections from the skin of the human eye lid. Two pictures from a series of 100, showing dermal Merkel cells (M) surrounded by nerve terminals (T) and terminal glial cells (S). Cytoplasmic protrusions (asterisks) extend from the Merkel cells through the glial cell sheath into the dermis. K, keratinocytes. Scale bar = 5 μm in b (applies to a,b).


In ectoderm-derived mucosa (mouth, lips, and anal canal), Merkel cells are found in many mammalian species. In the mouth, three different types of mucosa can be distinguished. First, the so-called surrounding mucosa covers the surface of lips, cheeks, and the soft palate. It is not cornified and forms epithelial pegs apart from the soft palate. The bases of these pegs harbor numerous Merkel nerve endings. Their structure is similar to Merkel nerve endings in the skin. Second, the specialized mucosa covering the surface of the tongue is free of Merkel nerve endings and forms typically structured papillae with gustatory organs. The last type of mucosa withstands heavy mechanical loads during chewing of food (called masticatory mucosa). Masticatory mucosa covers the hard palate and forms the surface of the gingiva. On its surface, the epithelium is cornified. The basal layer of the epithelium forms rete ridges or rete pegs. Many Merkel nerve endings (Fig. 9) can be found in the basal layer of such ridges or pegs. Their structure is identical with the structure of Merkel nerve endings in glabrous skin. However, their orientation shows species-dependent differences. In rhesus monkey, they are arranged perpendicular to the mucosa surface (Halata and Baumann, 1999), whereas in goat, they are parallel (Halata et al., 1999), in line with the differences in mechanical stimuli experienced during feeding.

Figure 9.

Ultrathin cross-section through the mucosa of the papilla incisiva in rhesus monkey. Merkel cells (M) with cytoplasmic protrusions (asterisks) are seen in the basal layer. Desmosomal contacts with surrounding epithelial cells are marked by arrows. Nerve terminals (T) from the afferent axon (N) covered by a Schwann cell.

In the mucosa of the anal canal, Merkel nerve endings can be found in nonkeratinized and keratinized parts. In the nonkeratinized part, the epithelium is a flat stratified epithelium with rete pegs. In the bases of the rete pegs, typical Merkel nerve endings are present. Also the keratinized part shows rete pegs with numerous Merkel nerve endings (Rettig and Halata, 1990).


Structurally, Merkel cells resemble cells of the diffuse endocrine system that are mainly found in the epithelium of the gut and bronchial mucosa. They are referred to as cells of the APUD system (Pearse, 1968) or paraneurons (Fujita, 1977). These cells are believed to release neurotransmitters with paracrine functions. Also, in those areas where Merkel cells are usually found, pale oval cells with dense-core granules are sometimes seen in the basal layer of the epidermis or the mucosa of ectodermal origin. In contrast to normal Merkel cells, these cells have oval nuclei with many nuclear pores and lack any contact with nerve terminals, as verified by examination of serial sections (Fig. 10a,b). They have few small desmosomes with surrounding keratinocytes, or hemi-desmosomes with the basal lamina, but lack cytoplasmic processes. The dense-core granules are close to the Golgi apparatus in that part of the cell facing the skin surface, rather than the basal lamina.

Figure 10.

a,b: Ultrathin cross-section through the skin of the human eye lid. Clearly distinguishable from the keratinocytes (K) are pale cells (N) in the basal layer of the epidermis. In contrast to Merkel cells, the nuclei of these cells are not lobulated but oval with numerous nuclear pores. The cells have no contacts with nerve terminals, and the osmiophilic granules are in the apical part of the cytoplasm. F, fibroblast.

Tachibana observed in the oral mucosa of rodents and man “dendritic Merkel cells” lacking any contact with nerve terminals (Tachibana et al., 1997, 1998). The authors suggested that the population of Merkel cells is very heterogeneous, with a wide variety of functions, including endocrine functions (Tachibana, 1995). In our own studies examining the papilla incisiva of goat and monkey we did not observe “dendritic” Merkel cells (Halata and Baumann, 1999; Halata et al., 1999). Other investigators, examining the buccal pouch of the hamster, did find such “dendritic” Merkel cells, although these made contact with nerve terminals (Tazaki et al., 2000). Thus, it cannot be said with certainty whether we are dealing with different cell populations with possibly different origins, or whether, particularly in epithelia exposed to heavy mechanical load, the same type of cell may have different forms and functions throughout its lifetime.


A highly malignant type of skin tumor is characterized by cells with osmiophilic granules and a positive immunohistochemical reaction with neuron-specific enolase (Toker, 1972; Johannessen and Gould, 1980). Later, it was shown that cells of these tumors stain positively for neurofilaments (Gould et al., 1985) and low molecular cytokeratins (especially CK 20, Moll et al., 1992; Chan et al., 1997), also features of normal Merkel cells (Leonard et al., 1993; Moll et al., 1995; Hashimoto et al., 1998). Therefore, it is widely assumed that these tumors originate from Merkel cells, giving rise to the name “Merkel cell carcinoma.” However, direct evidence for this assumption is still lacking (Gould et al., 1985; Moll et al., 1992). The high malignancy of this relatively rare tumor triggered an interest of many dermatologists and pathologists in Merkel cells and Merkel cell carcinomas. As the histogenesis of this tumor is still uncertain, another assumption is that this tumor may originate from pluripotent stem cells of the dermis (Hoefler et al., 1985). Leonard and Bell (1997) examined cell phenotypes of several different Merkel cell carcinoma cell lines. They concluded that these tumors were derived from cells originating from the neural crest, because they expressed transcription factors seen only in neuronal tissue. Only some “classic” cell lines exhibited neuroendocrine markers, whereas variant lines often lost the expression of some markers (Leonard and Bell, 1997).

The location where most Merkel cell carcinomas are seen does not correspond with the areas where most Merkel nerve endings are found, e.g. Merkel cell carcinomas of the finger tips and palms are extremely rare. Therefore, it appears likely that Merkel cell carcinomas originate from those Merkel cells in the skin that do not contact nerve terminals, but instead belong to a diffuse endocrine system.


Both in birds and mammals, including man, Merkel cells can occasionally be found in other types of mechanoreceptors. In pigeon (Pac, 1982), quail, and chick/quail chimeras (Grim and Halata, 2000a), Merkel cells were found within the capsule as well as the subcapsular space of Herbst corpuscles—the Pacinian-like receptor in birds. In mammals, we found Merkel cells in direct contact with lanceolate nerve terminals in the connective tissue of velus and guard hairs (Fig. 11). They also occur in the genital end bulb mechanoreceptor (Fig. 12) of the human glans penis (Halata and Munger, 1986). Considering the neural crest origin of Merkel cells (see below), one would interpret these cases as representing misguided migration.

Figure 11.

Ultrathin cross-section through a hair follicle from the human thigh skin. Lanceolate nerve terminals (L) are in direct contact with the basal lamina of the hair follicle (H). Occasionally Merkel cells (M) are found in the dermis associated with lanceolate nerve terminals and surrounded by cytoplasmic processes of Schwann cells (S).

Figure 12.

Ultrathin cross-section through the skin of the human glans penis. A Merkel cell (M) with typical cytoplasmic protrusions (arrow) is associated with nerve terminals (T) from so-called genital bulbs (similar to Meissner corpuscles of primate glabrous skin). The thin flat processes separating the nerve terminals within the genital bulb are formed by the terminal Schwann cell (S).


Merkel nerve endings were initially described in the featherless skin of the beak and legs of birds (Saxod, 1978a; Halata and Grim, 1993). In contrast to mammals, Merkel cells here are found exclusively in the dermis (Fig. 13), and the avian epidermis does not contain any nerve endings (Hemming et al., 1994).

Figure 13.

Semithin cross-section through quail beak skin. In the dermis, below the epidermis (E), numerous Merkel cells (M) with nerve terminals are visible.

In the beak skin of chick and quail, clusters of Merkel cells are found in the dermis between the epithelial pegs (Fig. 13). These clusters can contain up to 50 Merkel cells, with nerve endings surrounded by an incomplete capsule formed by thin modified glial cells. Discoid nerve terminals with large numbers of mitochondria are found between these cells (Fig. 14). The cell membrane facing the capsule has thin protoplasmic processes extending into the cells of the capsule or into the connective tissue of the dermis, whereas the membrane between Merkel cell and nerve terminals shows rare synapse-like connections. Afferent axons of 3 to 5 μm in diameter can each supply up to six Merkel cells. After losing their myelin sheath, they branch several times before forming discoid nerve terminals with synaptic contacts onto Merkel cells. Osmiophilic granules, of approximately 60 nm in diameter, are concentrated in the cytoplasm adjacent to the nerve terminals.

Figure 14.

Ultrathin cross-section showing two twin-groups of Merkel nerve endings from the tarsometatarsal skin of quail. K, keratinocyte; M, Merkel cells with cytoplasmic protrusions (arrows), T, nerve terminals; S, terminal Schwann cells.

The tarsometatarsal and digital regions of the skin have keratin scales. Below these scales, Merkel nerve endings, twin groups (Fig. 14), as well as accumulations of up to 12 Merkel cells are found forming corpuscle-like structures in the superficial layer of the dermis. Their appearance is identical to those in the beak skin. In some cases, Merkel cells were seen without nerve terminal contact, and very rarely Merkel cells were even found between lamellae from Herbst corpuscles (Grim and Halata, 2000a).

Anseriform birds lack Merkel nerve endings, but commonly show Grandry corpuscles, which are regarded as modified Merkel nerve endings (Halata, 1971; Saxod, 1978b). In the Peking duck, they are located in the dermis approximately 50 μm below the epithelium and consist of small groups of two to five large oval or flat cells (diameter 30 to 50 μm), ultrastructurally resembling Merkel cells. Between these Grandry cells are discoid nerve terminals (Fig. 15). The cells are often symmetrically built with a round or oval nucleus in the centre, surrounded by cytoplasm with rough endoplasmic reticulum. The outer parts of the cytoplasm contain dispersed osmiophilic granules of approximately 60 nm diameter and bundles of microfilaments. On the surface of the Grandry cells, the part of the membrane facing the capsule forms thousands of microvilli invaginating the cells of the capsule and often making desmosome-like contacts with them. In contrast, the part of the Grandry cell membrane facing the nerve terminals is smooth. The number of nerve terminals varies with the size of the corpuscle, but they are usually all the terminals from one myelinated nerve fiber of approximately 6-μm diameter (having lost its myelin sheath on entering the corpuscle). The nerve terminals contain large numbers of mitochondria and some electron microscopically empty vesicles. Membrane thickenings can be seen along the contact area of Grandry cells and nerve terminals. Modified glial cells form the capsule of the Grandry corpuscle into which the microvilli from the Grandry cells are anchored. A basal lamina surrounds the outer surface of the cells, forming the capsule. In the mucosa of the tongue, slightly smaller Grandry corpuscles (of two or three cells) are found in the lamina propria just below the epithelium. They are often found in papillae of connective tissue bulging into the epithelium. Under the light microscope, this imitates an intraepithelial location. However, electron microscopically, they have all the characteristics of those in beak skin.

Figure 15.

Ultrathin section through a Grandry corpuscle. The corpuscle consists of Grandry cells (1) and discoid nerve terminals (2). A single axon (4) supplies all terminals. On entering the corpuscle the axon loses its myelin sheath. The corpuscle is surrounded by lamellae of terminal glial cells (3).


There is still controversy about the developmental lineage of Merkel cells. According to one view, they originate from the neural crest and migrate into the mammalian epidermis during the embryonic period. This opinion is supported by findings in fetal and newborn mammals (Breathnach and Robins, 1970; Hashimoto, 1972; Winkelmann, 1977; Breathnach, 1978; Halata, 1981). The classification of Merkel cells as part of the APUD system (Pearse, 1968) or paraneuron (Fujita, 1977), supports this assumption as both systems are considered to be neural crest derivatives. However, it has been shown that many neuroendocrine cells develop from other sources than the neural crest (Le Douarin and Teillet, 1973), and as such may be endodermal (Andrew, 1982) or mesodermal in origin (Forssmann et al., 1983).

An alternative view is that Merkel cells arise from a common ectodermal stem cell (Munger, 1965; English, 1974; Moll et al., 1986a; Moll and Moll, 1992). This hypothesis is mainly supported by the finding of low molecular cytokeratins (CK 8, CK 18, CK 19, and CK 20) in Merkel cells of mammals (Moll et al., 1984, 1986b, 1995; Kim and Holbrook, 1995). Also, the presence of desmosomal contacts between Merkel cells and keratinocytes is interpreted as support for the ectodermal origin hypothesis (Munger, 1965). Currently, most investigators favor this opinion.

However, with respect to the neural crest hypothesis, it should be remembered that the neural crest gives rise not only to neuronal and glial cells, but also to epithelial C-cells in the thyroid gland, melanocytes, smooth muscle cells, as well as cells of connective tissue, bone, and cartilage (Le Douarin and Kalcheim, 1999). All of these cell types possess a very broad spectrum of different cytoskeletal filaments.

Experimentally, the developmental origin of Merkel cells was tested in amphibians by Tweedle (1978); in birds by Saxod (1980), Halata et al. (1990), Grim and Halata (2000a); and in mammals by Moll et al. (1990b) and Szeder et al. (2002). To determine the developmental origin of Merkel cells, the chick/quail marking system (Le Douarin, 1973) was used in our experiments. Cells of the quail host that migrated into grafted chicken leg primordium were identified according to heterochromatin blocks associated with the nucleolus of quail cells, which are absent in cells of chick (Fig. 16a). Our results showed that avian Merkel cells do not develop from ectodermal or mesodermal cells of the leg primordium, but that they migrate from the neural crest–like Schwann cells and melanoblasts. In quail leg with the primordium grafted on the chicken host embryo (Fig. 16b), the origin of Merkel cells was determined in serial semi- and ultrathin sections showing the absence of the nucleolus-associated heterochromatin blocks (Le Douarin, 1973). Our experiments provided evidence that Merkel cells migrate together with progenitors of Schwann cells and melanoblasts into the leg primordium (Halata et al., 1990; Grim and Halata, 2000a, b). They enter the base of the leg bud from stage 19 (of the Hamburger and Hamilton method) onward, i.e., before the innervating nerves. Thus, Merkel cells in birds represent yet another subpopulation of neural crest-derived cells.

Figure 16.

a: Ultrathin section of a developing Merkel nerve ending in the dermis of a chimeric chick leg 14 days after grafting the leg bud onto a quail host embryo. The Merkel cell (M) is not yet fully differentiated and contains only few dense-core granules. The large heterochromatin mass associated with the nucleolus identifies the Merkel cell as originating from the quail host. The nerve terminals are indicated by arrows and are completely covered by cytoplasm from Schwann cells (S). b: Ultrathin section of a developing Merkel nerve ending in the dermis of chimeric quail leg at embryonic day 20. The quail leg primordium was grafted on a chick host embryo at embryonic day 3. The Merkel cell (M) was identified in serial sections as immigrated from the chick host because of absent heterochromatin block, in contrast to surrounding cells. The nerve terminals are indicated by arrows. Scale bar = 5 μm in a (applies to a,b).

Our results are in line with findings of Saxod (1980) using chick and duck chimeras. The results exclude the epidermal origin of avian Merkel cells, and support the view that Grandry cells (the equivalent of Merkel cells in anseriform birds) and avian Merkel cells are of neural crest origin. In contrast, Tweedle's experiments in amphibians were interpreted as support for the epidermal origin of Merkel cells (Tweedle, 1978). Tweedle found Merkel cells in aneurogenic forelimbs of Ambystoma maculatum larvae after presumptive neural tissue removal. However, the gill region was left intact and neural crest cells from the gill region could have migrated into the forelimbs. The removal of nervous tissue in Tweedle's experiments was checked by ultrastructural examination for the presence of nerves, but not for the presence of neural crest cells. It has been shown that defects in the neural crest can be compensated by neighboring segments (Vaglia and Hall, 1999). Thus, Tweedle's results do not unambiguously exclude the neural crest origin of Merkel cells in Ambystoma.

Moll et al. (1990b) xenografted skin from human fetuses onto the subcutis of nude mice and found (after 4 to 8 weeks) human Merkel cells in the xenografts. The authors interpreted this observation as differentiation of Merkel cells from epidermal precursor cells. The skin was removed from fetuses (gestation weeks 8 to 11), and at the time of grafting, neural crest cells could already be found in the graft because neural crest-derived melanocytes were observed in the epidermis of human embryos starting from week 6 of gestation (Holbrook et al., 1989). Moreover, subepidermal nerve plexuses in human embryonic skin containing Schwann cells were also observed at week 6 of gestation (Moore and Munger, 1989; Terenghi et al., 1993). At the time of grafting, Moll et al. (1990b) found no Merkel cells in the skin using immunohistochemical staining for CK 18. However, according to Narisawa et al. (1992), human Merkel cells did not show cytokeratin immunoreactivity before week 12, whereas in another study, Moll and Moll (1992) were able to demonstrate CK 20-positive Merkel cells in epidermis of human fetuses as early as during fetal week 8. Thus, Merkel cells could well have been in the skin graft but were probably not detected by the immunohistochemical methods used at the time. Hence, the experiments by Moll et al. (1990b) cannot exclude the neural crest origin of Merkel cells with any certainty either.

In recent years, it has been shown that two transcription factors seen predominantly in nervous tissue were also detected in Merkel cells. This finding supports the neural crest origin of Merkel cells in mammals. The basic helix-loop-helix transcription factor Math1 was detected in Merkel cells of mouse hairy skin, whiskers, and footpads (Ben-Arie et al., 2000; Helms et al., 2000). The Math1 gene is one of the mouse homologues of atonal, known to be important for the development of the peripheral nervous system in Drosophila. Mouse homologues of atonal are expressed during neurogenesis both in the central and in the peripheral nervous system, including the dorsal portion of the neural tube. More recently, it has been shown that Math1 is essential for the development of certain components of the proprioceptive pathway, inner ear hair cells, cerebellar granule cells, and pontine nuclei (Bermingham et al., 1999, 2001). The expression of Math1 in inner ear hair cells and Merkel cells suggests that both receptor cells are involved in the conversion of mechanical stimuli into neuronal signals. Surprisingly, Math1 expression has also been found in chondrocytes of developing joints, which do not have any neural function (Ben-Arie et al., 2000).

In addition, both Hath1 protein, the human analogue of Math1, and the transcription factor Brn-3c have been detected in Merkel cell carcinoma cell lines and also in normal human Merkel cells (Leonard and Bell, 1997; Leonard et al., 2002). Brn-3c belongs to the family of POU-domain transcription factors that has been detected during mouse embryogenesis in subsets of retinal ganglion cells, as well as in somatosensory neurons of the dorsal roots and cranial nerves, including inner ear sensory ganglia (Xiang et al., 1996). It is believed to play a role in the activation of specific programmes of gene expression that define specific cell phenotypes (Rosenfeld, 1991).

Although there had been clear evidence for a neural crest origin of Merkel cells in birds for some time, a report providing evidence for a neural crest origin of mammalian Merkel cells has been published most recently (Szeder et al., 2003). The authors examined the ontogenetic origin of Merkel cells in Wnt 1-cre/R26R double transgenic mice. In these mice, the transient expression of cre recombinase under the control of the Wnt 1 promoter specifically and permanently activates expression of R26R derived β-galactosidase in premigratory and early migratory neural crest cells (Chai et al., 2000; Jiang et al., 2002). All Merkel cells of whiskers in these double transgenic mice express β-galactosidase, identifying these cells as neural crest derivatives. Immunoelectron microscopy clearly shows peroxidase reaction product in β-galactosidase–immunostained Merkel cells. Thus, there is now unambiguous evidence that also Merkel cells in mammalian whiskers originate from the neural crest.


Merkel's original description of “touch corpuscles” left no doubt that he considered these complexes as mechanoreceptors (Merkel, 1875). There is ultrastructural evidence that Merkel cells have synaptic contacts with the associated nerve terminals from pseudounipolar neurons of craniospinal ganglia (Iggo and Muir, 1969; Chen et al., 1973). These complexes are known to serve as mechanoreceptors (Iggo and Findlater, 1984). They respond to punctate pressure on the skin and bending of hairs with long-lasting spike trains (Iggo and Muir, 1969; Gottschaldt et al., 1973). The function of a mechanoreceptor is to convert mechanical stimuli into nerve action potentials. In all locations, Merkel cells are positioned, relative to the nerve terminal, on the side that normally receives mechanical stimulation. Furthermore, the surface of Merkel cells is equipped with protoplasmic protrusions anchoring them between keratinocytes (Iggo and Muir, 1969; Halata and Baumann, 2000). This observation has led to comparisons with inner ear hair cells (Iggo and Findlater, 1984; Baumann et al., 1990), where bending of the hairs opens mechanically gated ion channels, resulting in depolarization, an increase in free intracellular calcium, and finally transmitter release (Crawford et al., 1991).

There has been a long-standing controversy over whether the mechanoelectric transduction process occurs in the Merkel cell or whether the Merkel cell only directs the mechanical stimulus toward the nerve terminal. Unfortunately, the location of these receptors makes direct electrophysiological recordings with microelectrodes impossible. The slowly adapting responses of these receptors and the characteristic impulse pattern of action potentials with rather irregular interspike intervals were interpreted as evidence of a synaptic link (Iggo and Muir, 1969; Horch et al., 1974). On the other hand, the high frequency of action potentials at which these receptors can follow sinusoidal stimuli has been taken as an argument against synaptic transmission (Gottschaldt and Vahle-Hinz, 1981). As mentioned in the section on “immunohistochemical reactions,” various potential neurotransmitters were found in the dense-core granules of Merkel cells (Hartschuh et al., 1979; Hartschuh and Weihe, 1988; Cheng Chew and Leung, 1991; for a detailed review, see English et al., 1992). However, attempts to impair transmission with the respective receptor blockers were unsuccessful (Gottschaldt and Vahle-Hinz, 1982).

Diamond's group argued that Merkel cell mechanoreceptors still function after selective destruction of Merkel cells after quinacrine loading and bleaching with ultraviolet light (Diamond et al., 1988; Mills and Diamond, 1995). There are contradictory reports concerning the effect of such procedures on the responses of Merkel cell mechanoreceptors (Ikeda et al., 1994; Senok et al., 1996). However, electronmicroscopic examination of such ultraviolet-irradiated Merkel cell nerve terminal complexes, show that the tissue damage is not selective but extends to the nerve terminals as well, while leaving some Merkel cells relatively unharmed (Senok et al., 1996). Thus, this experimental approach does not answer the question of where the mechanoelectric transduction process actually occurs.

Recently, dyes have become available that are taken up by living cells and emit fluorescent light, as such demonstrating the free intracellular calcium level (Grynkiewicz et al., 1985). These dyes allow examination of the intracellular calcium concentration in Merkel cells during mechanical stimulation and provide significant results about the function of Merkel cells in the mechanotransduction process. There are reports of increases in intracellular calcium concentration in Merkel cells during mechanical stimulation in different species and locations of Merkel cell receptors (Chan et al., 1996; Tazaki and Suzuki, 1998). A consistent increase in free intracellular calcium of Merkel cells can be observed during direct mechanical stimulation of Merkel cells with microprobes (Fig. 17), as well as during slight swelling of Merkel cells produced by exposure to hyposmotic solutions (Fig. 18). Despite mechanical stimulation, no such increase is seen if the extracellular fluid is calcium free (Fig. 18a) or if amiloride (known to block mechanosensitive ion channels; Hamill et al., 1992) is added to the solution (Fig. 18b). Thus, mechanical stimulation causes an influx of calcium into Merkel cells, which in turn appears to trigger further release of calcium from intracellular stores (calcium-induced calcium release; Senok and Baumann, 1997). It has long been established that an increase in free calcium is required for synaptic transmitter release (Katz and Miledi, 1967). In addition, there is recent experimental evidence showing that glutamatergic transmission occurs at Merkel cell receptors (Fagan and Cahusac, 2001) and is probably essential for the slowly adapting responses during the static phase of mechanical stimuli (Ogawa, 1996). The type of glutamate receptor involved has not yet been identified.

Figure 17.

Effect of mechanical stimulation on cytosolic calcium in Fluo-3-loaded Merkel cells. Rat sinus hairs are dissected while continuously superfused with oxygenated synthetic interstitial fluid (SIF-Bretag, 1969). The blood sinus is carefully opened exposing Merkel cells only covered by the glassy membrane (see Fig. 6b,c). Fluo-3/AM (2 μM) is added to the SIF for 1 h, allowing diffusion of the esterified dye into Merkel cells. This step is followed by washing off the incompletely de-esterified forms of the dye by superfusion with normal SIF for 45 min. Then, the sinus hair is transferred to a microfluorimetric setup, consisting of a Nikon Diaphot TMD-EF inverted microscope equipped with photomultipliers (Newcastle Photometric Systems). By using an excitation wavelength of 490 nm, the emitted light of 520 nm is proportional to the free intracellular calcium concentration (for further details, see Chan et al., 1996). A micromanipulator driven glass rod was carefully pushed against Merkel cells and kept in position during the time indicated by the bar at the top. (Mean ± SEM of seven Merkel cells.)

Figure 18.

a: Effect of hyposmotic swelling on cytosolic calcium in Merkel cells using confocal microfluorimetry. Absence of calcium in the extracellular fluid abolishes the intracellular calcium increase, suggesting an influx of calcium through the membrane. (Mean ± SEM of eight Merkel cells.) b: Effect of hyposmotic swelling on cytosolic calcium in Merkel cells using confocal microfluorimetry. Addition of 1 mM amiloride in the bathing solution almost abolishes the intracellular calcium increase (Mean ± SEM of seven Merkel cells). Sinus hairs were prepared as outlined in the legend of Figure 17. However, mechanical stimuli were not applied to Merkel cells directly but by means of superfusion with hyposmotic SIF (95% of normal through omission of gluconate and sucrose), resulting in slight swelling of cells. The resulting increase in free intracellular calcium concentration was again measured through Fluo-3 fluorescence by using a BioRad MRC1000 confocal imaging system.


Neuroendocrine cells are characterized by the occurrence of biogenic amines and peptides. Similar substances were found in the osmiophilic granules of Merkel cells, leading to the classification of Merkel cells as part of the family of the APUD system (Pearse, 1968; Winkelmann, 1977). It was assumed that these biogenic amines and peptides might function as neurotransmitters at the putative synapse between Merkel cell and nerve terminal (Hartschuh et al., 1979). However, neither met-enkephalin nor any of the other substances tested, had significant effects on receptor responses to mechanical stimuli (Gottschaldt and Vahle-Hinz, 1982).

Ignoring the mechanoreceptive function of Merkel cells, some authors consider them neuroendocrine cells of the skin (Boot et al., 1992) and mucosa (Tachibana, 1995), with paracrine functions (Hartschuh and Weihe, 1989). This assumption was supported by the observation of an increased number of CK 18-positive cells in sun exposed skin (Moll et al., 1990a). However, evidence for exocytosis of substances from Merkel cells supporting the assumption of a paracrine (Fujita, 1977) or neuroendocrine role is still lacking. As such, the question about the functional role of those substances in the dense-core granules of Merkel cells still remains unanswered.

The occurrence of isolated Merkel cell-like cells without association with a nerve terminal has lent support to the theory of a neuroendocrine function. However, the number of cutaneous Merkel cells without nerve fiber contact is rather small. Pasche et al. (1990) studied the relationship of Merkel cells with nerve endings during embryogenesis in the mouse epidermis. They counted several thousand Merkel cells and found that already on embryonic day 17, approximately 95% of Merkel cells in whisker pad, and 85% in the back skin, were innervated. To the best of our knowledge there are no other quantitative studies investigating the number of Merkel cells with and without associated nerve terminals in postnatal mammals.


Merkel described in 1875 large, pale cells (called “Merkel cells”) in close contact with enlarged terminal branches of myelinated afferent nerve fibers (“Merkel disks”). Later, it was confirmed that these structures function as mechanoreceptors in various types of skin in all vertebrates. Merkel cells are best identified at the electronmicroscopic level through characteristic features (lobulated nucleus, finger-like protoplasmic protrusions in the part of the cell opposite to the nerve terminal, and dense-core granules in the part of the cytoplasm facing the nerve terminal). For light microscopic identification, low molecular cytokeratins, especially CK 20, are highly specific Merkel cell markers in the skin. However, other epithelial cell types also show positive reaction for CK 20. Consequently, not every CK 20-positive cell is a Merkel cell.

Occasionally, cells showing the above-mentioned features but without contact to nerve terminals are seen in mammalian epidermis and dermis. These cells do not function as mechanoreceptors and are believed to belong to a diffuse neuroendocrine system. Clear information on their function is still lacking, and it is still unknown whether these cells belong to the same developmental lineage as those Merkel cells in contact with nerve terminals.

The origin of Merkel cells has been the subject of opposing views in the literature. Although some authors argue that Merkel cells originate from epidermal cells, other studies support the view that they are derived from the neural crest. The latter has been proven beyond doubt in birds. Most recently, also in mammals, conclusive evidence for a neural crest origin of Merkel cells has been obtained.

A rare type of carcinoma in the skin is characterized by tumor cells with positive staining for CK 20 and osmiophilic granules. This finding has led to the term Merkel cell carcinoma. However, there is no evidence yet that these tumors originate from normal Merkel cells (with or without adjacent nerve terminals).

Although Merkel nerve endings are clearly identified as slowly adapting mechanoreceptors in mammals, there has been controversy about the site of the mechanoelectric transduction process. Although the high frequency of action potentials at which these receptors can follow sinusoidal stimuli has been taken as an argument against synaptic transmission, there is now unequivocal evidence for a mechanically induced increase in free intracellular calcium in Merkel cells, which is a prerequisite of any synaptic transmission. It appears now that glutamate is the most likely neurotransmitter, although the specific glutamate receptor still needs to be identified.


The authors thank Ms. Brigitte Asmus, Hamburg, Germany, and Ms. Eva Kluzakova, Prague, Czech Republic, for excellent technical assistance. The authors also thank Dr. Peter Cahusac, Stirling, United Kingdom, for helpful comments on the manuscript.