ARTICLE
Peripherin (PR) is a type III intermediate filament of 56-58 kDa MW [1-3],
encoded for by a gene located in humans on chromosome 12 [4]. It is expressed
in motor, sensory and sympathetic system neurons and appears during neurogenesis
after the expression of nestin, vimentin, alpha-internexin and neurofilaments
[5]. It may thus be considered as a "mature" neuronal cell intermediate
filament (IF), playing a role in the differentiation of neurons, namely
of the peripheral nervous system, and also probably in axonal regeneration
[2]. Intermediate filaments have long been used as immunohistochemical
markers for the diagnosis of tumors since their expression is usually
maintained during the neoplastic transformation of cells; they can therefore
be used for the determination of the histogenetic origin of morphologically
undifferentiated neoplastic proliferations [6]. However PR has so far
attracted little attention in dermatopathology. Some years ago it was
reported that neuroendocrine skin carcinomas express PR [7], and more
recently PR was claimed to be expressed in benign and malignant melanocytic
tumors [8, 9]. The purpose of this study was to investigate the potential
usefulness of antibodies to PR in the diagnosis of skin tumors, with emphasis
on those of neural origin. The expression of PR was studied immunohistochemically
in a wider panel of skin tumors and was compared with the expression of
neurofilaments, the other type of IF known to be specific to nerve fibres
and neural tissue tumors.
Materials and methods
Tissue samples. These included 68 cutaneous tumors collected
in our dermatopathology laboratory over the past three years; they included
15 neurofibromas, 4 schwannomas, 5 neuromas, 2 granular-cell tumors (Abrikossoff),
7 dermal (neurotized) nevi, 5 Spitz nevi, 6 blue nevi, 19 melanomas (15
primary, 4 metastatic) and 5 neuroendocrine skin carcinomas ("Merkel-cell
tumors") (Table I). The
material had been fixed in formalin and embedded in paraffin. The diagnosis
had been made upon examination of hematoxylin-eosin-stained sections,
according to standard histological criteria, supplemented (when needed)
by appropriate immunohistochemical staining.
Immunohistochemistry. Four µm-thick sections placed on clean,
positively-charged glass slides were deparaffinized and rehydrated, then
immunolabeled using a streptavidin-biotin-peroxidase technique (kit LSAB
Dako, Copenhagen, Denmark) after antigen retrieval (heating the slides
immersed in citrate buffer in a microwave oven 2 x 3 min). The
immunolabeling protocol included the following steps: a) inhibition of
endogenous peroxidase with 1% H2O2 in PBS; b) incubation
of the sections with blocking (non-immune) serum; c) incubation with primary
antibodies (see below); d) incubation with biotin-conjugated antiserum
to mouse immunoglobulins (10 min); e) incubation with peroxidase-conjugated
streptavidin (10 min). The reaction was revealed with aminoethylcarbazole
as chromogen. Primary antibodies included: a) a rabbit polyclonal antiserum
to peripherin (Chemicon Int, Temecula, CA), produced against electrophoretically
pure trp-E-peripherin fusion protein containing all but the four N-terminal
amino acids of rat peripherin; this reagent, recognizing rat, mouse, human,
pig and cow peripherin, was applied at a dilution of 1:100 for 15 min
in moist chambers at room temperature; and b) a pooled antibody made up
of equal ratios of monoclonal antibodies to 70, 160 and 210 kDa neurofilaments,
i.e. NF-L, NF-M and NF-H (Immunotech, Marseille, France). This
was applied at a dilution of 1:20 for 15 min in moist chambers at room
temperature. Negative controls were performed by omitting the first layer
antibody and proved consistently negative.
Results
In normal skin and in the skin adjacent to the tumors studied, PR immunoreactivity
was detected as fine, granular staining within the cytoplasm of axons
of dermal nerves. No labeling was seen within the epidermis. Compared
with NF, PR showed an identical cellular distribution; however, staining
for NF usually had a coarser aspect, corresponding to the entire cytoplasm
of axons, whereas PR expression was more delicate, often assuming a granular
or dot-like pattern.
In the lesions studied, axons of dermal nerves observed within and around
tumor cells expressed both PR and NF and served as built-in controls.
Tumors of neural origin, such as neurofibromas, schwannomas and granular-cell
tumors (schwannomas) contained a variable number of axons, whose number
was maximal in neuromas; however, the proliferative tumor cells themselves
were unreactive for both PR and NF (Figs.
1 and 2).
The majority (15/18, i.e. 83%) of benign nevi (including 100% of
neurotized nevi) showed no specific labeling of tumor cells; two out of
six blue nevi and one out of five Spitz nevi contained a small percentage
of cells (ca. 10 and 20%, respectively) expressing weak cytoplasmic
PR (but not NF) immunoreactivity. Three out of six SSM and four out of
eight nodular malignant melanomas showed a small percentage (ca.
10%) of PR-immunoreactive cells; these lay either within the basal epidermal
layer, or within the papillary dermis (Fig.
3), deep-seated melanoma cells being unreactive. Two and four
(out of five) neuroendocrine skin carcinomas (NSC) expressed cytoplasmic
PR and NF immunoreactivity, respectively, in a paranuclear, dot-like pattern.
Three NSC showed a higher reactivity for NF in terms of percentage of
immunoreactive cells but one tumor showed a clearly predominant PR immunoreactivity
as compared with NF.
Discussion
Peripherin (PR) has been shown to be expressed by a number of neural
crest tumors, including (ganglio)neuroblastomas, ganglioneuromas and gangliogliomas
[10]. Its expression has also been used for the study of malformations
of the enteric nervous system, where it was found to be a superior marker
for the recognition of ganglion cells compared with other markers (neuron-specific
enolase, S100 protein, NF and synaptophysin) [11]. Few data exist in the
literature concerning PR expression in normal and diseased skin, aside
from four studies reporting PR immunoreactivity in neuroendocrine carcinomas
of the skin [7, 12], melanocytic [8] and neural tumors [9]. Our results
show that in normal adult human skin PR has the same overall cellular
distribution as NF, being expressed within dermal neurons. We found no
expression of PR in Merkel cells and melanocytes of the epidermis or its
adnexae, considered to be of neural crest origin, in keeping with the
results of previous studies [8, 9, 12]. Although some studies have reported
NF expression within epidermal Merkel cells, we did not detect such a
reactivity, in keeping with the majority of studies reported so far [13].
Also in the neural tumors studied (neurofibromas, ordinary and granular
cell schwannomas, neuromas), PR expression closely paralleled that of
NF, being observed within axons entrapped within the tumor masses but
not within tumor cells themselves. This finding is in keeping with the
results of a previous study [9] and with the admitted cellular origin
of these lesions from Schwann cells and perineural fibroblasts (rather
than neurons). In neuroendocrine carcinomas of the skin we found specific
labeling for PR in a variable percentage (20-70%) of tumor cells in three
cases, that was lower of that for NF (10-90%) in three out of the four
cases showing either PR or NF reactivity. The positivity rate we found
(2/5 or 40%) for PR ranged between those found in two previous studies,
i.e. 2/8 or 25% [12] and 10/12 or 83% [7].
Contrasting with the results of Prieto et
al. [8], who found PR expression in the great majority of benign nevi
and melanomas, we found only limited PR expression in this tumor group.
Indeed, we found only 2/6 blue nevi and 1/5 Spitz nevi showing weak PR
immunoreactivity; similarly, most (12/19, i.e. 63%) malignant melanomas
did not express PR, with only 37% of them showing weak cytoplasmic PR
immunoreactivity; even in these positive cases, immunoreactivity was seen
in a small percentage of cells (less than 20%).
The significance of PR expression in melanocytic tumors is unclear.
Melanocytes are considered to derive from the neuroectoderm and the neural
crest, a fact consistent with the occasional expression of PR in tumors
arising there. However, normal melanocytes and the great majority of nevus
cells do not express PR; it can be speculated that PR expression could
be turned on by local transcription factors. Such candidate neuropoietic
factors include NGF, LIF (leukemia inhibitory factor) and IL-6, that have
been shown under experimental conditions to induce PR expression [14,
15]. Interestingly, from the present results and those of Prieto et
al. [8], it seems that PR-expressing cells of melanocytic lineage
are located within the upper dermis or the epidermis, as if epidermal
cells were responsible for this upregulation.
In conclusion, the expression of PR both in normal and neoplastic adult
human skin is restricted to dermal neurons and parallels the expression
of NF. Antibodies to PR can be useful for the demonstration of the neuronal
component of skin tumors in the same manner as antibodies to NF. Although
melanocytic and neuroendocrine tumors may express PR, the sensitivity
of PR for these tumors is lower than that of other established markers,
such as NF, S100 protein or HMB45 antigen [6], making the diagnostic usefulness
of antibodies to PR in dermatopathology rather limited.
This work was presented at the International Investigative Dermatology
Meeting, Köln, Germany (7-10/5/98).
CONCLUSION
Acknowledgements
We are indebted to J. Soum and G. Paret for technical assistance.
REFERENCES
1. Portier MM, de Nechaud B, Gros F. Peripherin, a new member of the
intermediate protein family. Dev Neurosci 1983-1984; 6: 335-44.
2. Ho CL, Liem R. Intermediate filaments in the nervous system: implications
in cancer. Cancer Metast Rev 1996; 15: 483-97.
3. Bousquet O, Portier MM. Les protéines de filaments intermédiaires
neuronales. CR Soc Biol 1996; 190: 269-87.
4. Moncla A, Landon F, Mattei M, Portier M. Chromosomal localization
of the mouse and human peripherin genes. Genet Res 1992; 59: 125-9.
5. Nixon R, Shea T. Dynamics of neuronal intermediate filaments: a developmental
perspective. Cell Motil Cytoskel 1992; 22: 81-91.
6. Kanitakis J. Solid cutaneous tumors. In: Kanitakis J, Vassileva S,
Woodley D, eds. Diagnostic Immunohistochemistry of the Skin. London,
Chapman & Hall Med., 1998: 279-99.
7. Baudoin C, Meneguzzi G, Portier M, Demarchez M, Bernerd F, Pisani
A, Ortonne JP. Peripherin, a neuronal intermediate filament, is stably
expressed by neuroendocrine carcinomas of the skin, their xenograft on
nude mice, and the corresponding primary cultures. Cancer Res 1993;
53: 1175-81.
8. Prieto V, McNutt S, Lugo J, Reed J. The intermediate filament peripherin
is expressed in cutaneous melanocytic lesions. J Cutan Pathol 1997;
24: 145-50.
9. Prieto V, McNutt S, Lugo J, Reed J. Differential expression of the
intermediate filament peripherin in cutaneous neural lesions and neurotized
melanocytic nevi. Am J Surg Pathol 1997; 21: 1450-4.
10. Foley J, Witte D, Chiu F, Parysek L. Expression of the neural intermediate
filament proteins peripherin and neurofilament-66/alpha internexin in
neuroblastoma. Lab Invest 1994; 71: 193-9.
11. Szabolcs M, Visser J, Shelanski M, O'Toole K. Peripherin: a novel
marker for the immunohistochemical study of malformations of the enteric
nervous system. Ped Pathol Lab Med 1996; 16: 51-70.
12. Alvarez-Gago T, Bullon MM, Rivera F, Velasco A, Mayo A. Intermediate
filament aggregates in mitoses of primary cutaneous neuroendocrine (Merkel-cell)
carcinoma. Histopathol 1996; 28: 349-55.
13. Moll I, Kuhn C, Moll R. Cytokeratin 20 is a general marker of cutaneous
Merkel cells while certain neuronal proteins are absent. J Invest
Dermatol 1995; 104: 910-5.
14. Aleta J, Angeletti R, Liem R, Purcell C, Shelanski M, Greene L.
Relationship between the nerve growth factor-regulated clone 73 gene product
and the 58 kDa neuronal intermediate filament protein (peripherin). J
Neurochem 1988; 51: 1317-20.
15. Lecompte M, Basseville M, Landon F, Karpov V, Fauquet M. Transcriptional
activation of the mouse peripherin gene by leukemia inhibitory factor:
involvement of STAT proteins. J Neurochem 1998; 70: 971-82.
|
Printable version
