Immunocytochemistry of cells of normal human · Glial fibrillary acidic protein and S100have been shownin the ... Of the six recognisable cell types within the human pituitary gland,

J Clin Pathol 1985;38:481-488

Immunocytochemistry of folliculo-stellate cells ofnormal and neoplastic human pituitary glandCHRISTINE S MORRIS, EDWARD HITCHCOCK

From the Brain Tumour Laboratory, Department ofNeurosurgery, University ofBirmingham, Birmingham

SUMMARY Five normal human pituitaries and 20 pituitary neoplasms were investigated byimmunocytochemical methods. Glial fibrillary acidic protein and S100 have been shown in theanterior lobe of the pituitary. Both these markers were present in the folliculo-stellate cell.Evidence is presented for the presence of a transitional folliculo-stellate cell which isimmunoreactive for S100. The role of the folliculo-stellate cell is discussed.

The characterisation of hormone secreting cellswithin pituitary neoplasms by immunocytochemicalmethods' is well established and has contributed toour understanding of the biology of such tumours. Itis surprising that exploration of the characteristics ofthese cells in response to neural or glial cell markershas received so little attention.Of the six recognisable cell types within the

human pituitary gland, five are hormone secretoryand the sixth is a less well defined cell type variouslyreferred to as a follicular cell,23 a stellate cell, anagranular stellate cell,' and folliculo-stellate cell.56About 3-4% of the total pituitary cell populationcomprises stellate cells with one or more processespassing to the perivascular space and to the centralpituitary follicle.2 Bergland and Torack2 havedescribed the electron microscopical appearance ofthese structures in detail. Fakuda4 points out thatthe cell changes its properties throughout life: infetal stages it is follicular but it progressively dif-ferentiates into a stellate form. On these grounds herefers to them as agranular stellate cells. The termfolliculo-stellate cell, however, appears to be thepreferable term.5

Although glial fibrillary acidic protein (GFAP) isregarded as a characteristic and specific glial marker,it is also present in the pituitary gland: it has beenfound in large amounts in rat pars nervosa9 '0 and,recently, it has also been shown in human pars dis-talis.8 The presence of GFAP in pars distalis can beattributed to cells originating from the neural tube,unlike the hormone secreting cells which originatefrom the epithelium of the stomatodeum. Two other

Accepted for publication 10 January 1985

markers, S100 and neurone specific enolase (NSE),have been extensively studied in relation to purelynervous structures. S100 is found in glial cells andalso in neurones and has been shown in the pituitarygland of the adult rat.5 There it has been found instellate cells of the pars distalis and tuberalis, in themarginal cells lining the hypophyseal cleft, and inthe glial like pituicytes of the neural lobe. It hastherefore been suggested that all these cells are ofneuroectodermal origin. S100, however, has a widerdistribution and has been shown in melanocytes ofthe skin and sweat glands, salivary glands, and thebreast and a number of other non-nervous neo-plasms such as carcinomas of the lung and teratomasof the ovary.'2 Thus a wide range of tissues containS100. Although a predominantly neuroectodermalmarker, its presence in a wide range of other tissuesmakes it a less valuable determinant than GFAP.NSE has been found in a number of neural

tumours including neuroblastomas,'3 '4 oligodendro-gliomas, astrocytomas, ependymomas, acousticneuromas, choroid plexus papillomas, and pituitaryadenomas.'5 It has also been found in other tumoursof non-neuroectodermal origin such as carcinoma ofthe breast and chordoma. Its major use has been as amarker of the amine percursor uptake and decar-boxylation (APUD) system.'4 16There is good evidence for the presence of these

three common neural markers within the pituitarygland, which we have investigated in otherimmunocytochemical studies on the pituitary.

Material and methods

The material comprised five normal humanpituitaries and 20 pituitary neoplasms. These were

481

on May 13, 2021 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.38.5.481 on 1 May 1985. D

ownloaded from

http://jcp.bmj.com/

482

all fixed in formalin, embedded in paraffin, and ser-ially sectioned. Immunocytochemistry was per-formed by the peroxidase-antiperoxidase (PAP)immunoperoxidase technique." 18

Commercially available antisera (Dako, MerciaBrocades, Byfleet, Surrey) were used for the GFAP,S100, and NSE determinations. The antisera sup-plied were of high titre and after titration were usedat their optimal dilutions of 1/300 for anti-GFAP,1/400 for anti-S100, and 1/500 for anti-NSE.

Preliminary studies showed that the results weremore reliable and reproducible when the tissue sec-tions were pretreated with trypsin. Normalpituitaries obtained at necropsy or surgical speci-mens of pituitary neoplasms were often heavily con-taminated by blood and it was essential to blockendogenous peroxidase. The optimal incubationtimes in trypsin and hydrogen peroxide solutionswere determined empirically.The PAP immunoperoxidase technique was car-

ried out as follows: the sections were deparaffinisedin three changes of xylene and ethanol and thenrinsed in two changes of phosphate buffered salinefor 5 min each before trypsinisation. All sectionswere pretreated with trypsin (0.1% trypsin in 0-1%Ca Cl2 solution, pH 7.8) for 20 min at 37°C. Theywere subsequently rinsed three times for 5 min eachin PBS containing 0-2% Triton X-100. All subse-quent rinsing was similarly performed. Endogenousperoxidase was blocked by incubation in 3% aque-ous hydrogen peroxide for 15 min at room tempera-ture. After rinsing the sections for 5 min in distilledwater and in two changes of PBS for 5 min each, astandard PAP technique was followed. The final

Fig. 1 Normal pituitary. Glial fibrillary acidic proteinpositive folliculo-stellate cells lining a Rathke's cyst. x300.

Morris, Hitchcock

immunoreactive product was visualised with3-amino-9-ethylcarbazole. The sections were coun-terstained with Celestin blue and Mayer'shaematoxylin and then mounted in glycerol gelatin(Sigma Chemical Company, Poole, Dorset, productcode GG-1).The following controls were included in all assays:

endogenous peroxidase control; substitution ofprimary antiserum with normal non-immune serum;omission of the bridge antiserum; omission of thePAP complex. All negative controls failed to showany immunoreactivity. In addition, positive controlsof suitable normal tissues were included for eachantiserum.The sections were then viewed by each author

independently and graded according to intensity ofstaining on a four point scale; the regions of maxi-mum staining were noted. Folliculo-stellate cellswere identified on the basis of morphology and dis-tribution. The England Finder* was used to identifyparticular cells and distinguish their individual stain-

*A stage graticule with reference guide permitting repositioning ofany standard (76 x 26 mm) slide allowing re-examination of agiven field. Carefully matched serial sections can thus be aligned onconsecutive slides permitting location of identical fields of interest.

-f

Fig. 2 Normal pituitary. Glial fibrillary acidic proteinpositive folliculo-stellate cells in pars distalis. Stellate formwith several processes radiating between granulated cells.xSOO.



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

Immunocytochemistry offolliculo-stellate cells of normal and neoplastic human pituitary gland

Fig. 3 Normal pituitary. SI00positive folliculo-stellate cells in parsdistalis in close association withhormone secreting cells. xSOO.

ing properties, and comparisons were made inselected areas and occasionally in selected cells.

Results

In general, normal human pituitary gland and pituit-ary neoplasms showed strong reactivity to all threeagents. There was a greater proportion of GFAPpositive cells than S100 positive cells in normalpituitary, but the reverse was true in pituitary neo-plasms. Pars nervosa, intermedia, and distalis werecarefully examined in the normal pituitaries and inpituitary tumours; a general survey of the whole sec-tion was made.

NORMAL PITUITARYPars nervosa showed strong staining for NSE, a les-ser amount for S100, and sparse GFAP reactivity. Inthe residual Rathke's cysts a variety of reactionswere seen (Fig. 1). Some cysts were lined with cellswhich were either GFAP or S100 or NSE positive,but many were negative. Occasionally, cysts wereseen lined completely by reactive cells. More usuallyboth positive and negative cells were present aroundthe cysts.GFAP positive folliculo-stellate cells were

scattered throughout the pars distalis, but theytended to occur in greater numbers closer to thecapsule, where they often stained intensely. Withinthe substance of the pars distalis their locationappeared to be more within the stroma than withinthe acinar structures. The cells were often pearshaped, but frequently had a distinct stellate formwith four to five processes with broad bases (Fig. 2).In some sections we were able to see these prolonga-tions continue between secretory cells with anappearance almost of investment. Other cells werespindle shaped and then were often seen close tocapillaries. These cells were compared with thehormone secreting cells and in no instance were theyrecognisably the same and appeared morphologi-cally quite distinct.S100 positive folliculo-stellate cells were scat-

tered throughout the pars distalis, usually in smallgroups of between five and ten cells and oftenclosely associated with hormone secreting cells (Fig.3). In stromal areas there was also staining similar tothat shown by GFAP positive cells.NSE positive cells were scattered throughout the

gland but again tended to concentrate closer to thecapsule. They differed from folliculo-stellate cells,however, in being large, plump, and granular and

483



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

484

corresponded with the corticotrophs (Fig. 4). Not allgranular cells were stained and it is probable that theremainder were accounted for by gonadotrophs, lac-totrophs, thyrotrophs, and somatotrophs.

Preliminary attempts at double labelling forGFAP plus S100, GFAP plus NSE, and S100 plusNSE supported the hypothesis that there may bemore than one population of folliculo-stellate cells.Some cells were GFAP positive, others S100 posi-tive, and a small population appeared to be reactivefor both. NSE reactivity was restricted to granulatedcells morphologically identical to hormone contain-ing cells, and none of the folliculo-stellate cells wasdouble labelled.

PITUITARY TUMOURSThese tumours comprised a variety of hormonetypes and included prolactinomas, bi-hormonal andmulti-hormonal tumours.GFAP positive folliculo-stellate cells were less

prominent in tumour tissue than in normal pituitary.They were scattered throughout the tumour masswith little tendency to accumulate in any site otherthan in the stroma (Fig. 5). In some instances theyjuxtaposed and interdigitated with the hormone

Morris, Hitchcock

producing cells. There was great variation in thenumber of GFAP positive folliculo-stellate cells fordifferent tumours, but there appeared to be no cor-relation between this variation and hormone type.In some tumours the GFAP positive cells were dif-fusely scattered through the tumour mass, with con-siderable variation in staining intensity between dif-ferent tumours and different sites within the sametumour. It was noticeable that these cells stained lessintensely when closely related to secretory cells. Incontrast, GFAP positive cells within the stromaoften showed intense staining. Similar intense reac-tivity could often be seen in areas immediatelybelow the capsule. Folliculo-stellate cells appearedto assume a more stellate shape when located closeto the secretory cells, compared with their appear-ance in stromal areas. Occasionally, they werelocated around microcysts and capillaries.

In general more cells were S100 positive thanGFAP positive. S100 positive cells showed a similarvariation in reactivity and distribution to GFAPpositive cells and, similarly, no correlation withhormone type. S100 positive cells, however, weremore profuse and more intensely staining thanGFAP positive cells and appeared in larger groups

Fig. 4 Normal pituitary. Neuronespecific enolase positive cells inpars distalis. Their morphology

I resembles that ofgranulatedhormone secreting cells and theyare seen in large groups. x500.



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

Immunocytochemistry offolliculo-stellate cells of normal and neoplastic human pituitary gland

Fig. 5 Pituitary adenoma. Glialfibrillary acidic protein positivefolliculo-stellate cells scatteredthroughout the stroma (darkstaining). x300.

,.I.AK. ...

t.^*?$.

.00M$8 j '

(Fig. 6). These cells seemed to have a more intimaterelation with secretory cells and interdigitationswere particularly prominent. There was no differ-ence in staining intensity between cells within thestroma and those among secretory cells. They werecommonly associated with microcysts and follicularstructures (Fig. 7).NSE positive cells were often numerous and cor-

related with the type of hormone secreting cells dis-tinguished by other means and were clearly notfolliculo-stellate cells. Double labelling with NSEand S100 and with NSE and GFAP in one tumourshowed that none of the folliculo-stellate cells couldbe double labelled; this confirms the results for thenormal pituitary.

Conclusions and discussion

This investigation has clearly shown a pituitary cellwhich is not hormone secreting but which bears aclose relation to those cells with which it often inter-digitates. Similar cells have been found in animalsand man and identified as folliculo-stellate cells. Todate, S100 positive cells have been found only in ratpituitary.5 6'1 GFAP positive pituitary cells havebeen found in both rat9'0 and man,8 but primarilythese were pituicytes of the neural lobe. There isonly one reference to the demonstration of GFAP inthe anterior lobe of man8 and none has reportedS 100. This investigation has shown that the

folliculo-stellate cells stain for GFAP and S100,although not all cells stained for S100 are GFAPpositive.

Cocchia and Miani" showed S100 protein withinthe stellate cells of the pars distalis and tuberalis of

^E ia ow Ji~~~~-~

S g sx§~~~~~~~~~,"

,_ * ~~~~~~~~. Xss..

S

cN0

<A.

:1'' ..4 eV ix

'p ;t<~~~~A'U¾X

Fig. 6 Pituitary adenoma. SI00 positive folliculo-stellatecells in clusters throughout the tumour. x125.

485

11

"VP.1

, . .0or lj.

t. ..

VW)k t

%sI

w

et. .' ,Aof -



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

Mornis, Hitchcock

Fig. 7 Pituitary adenoma. SI00positive folliculo-stellate cells closelyassociated with microcysts andfollicular structures. xsOO.

the rat and also in the hypophyseal cleft and thepituicytes of the neural lobe. The latter are regardedas glial like cells. These authors pointed to the com-mon properties of these cells, which are satellitecells to the secretory axons of the neural lobe andhormone secretory cells of the anterior lobe. Theytherefore suggest that all S100 positive cells in thepituitary are of neuroectodermal origin and gliallike. They note that all pituicytes contain S100 andthe protein distribution is the same in all cells.Nakajima et als also showed S100 protein within

the nucleus and cytoplasm of folliculo-stellate cellsand agreed that they were probably of neuroecto-dermal origin. They particularly noted the extensionof cytoplasmic processes between granular cells, acharacteristic feature which we have also observed.

Speculating on the role of these cells, Yoshimuraet al'9 considered the folliculo-stellate cell to berelated to the renewal cell system of the rat anteriorpituitary. Ohtsuka et a!20 cultured rat folliculo-stellate cells with corticotrophin releasing hormoneand suggested that they differentiated into bothacidophils and basophils. There is clearly a relationbetween folliculo-stellate cells and hormone secret-ing cells. Shirasawa et a!6 found that castrated ratsdeveloped numerous folliculo-stellate cells withcytoplasmic processes surrounding the gonado-trophs. They failed to show any topographic affinityfor thyrotrophs after thyroidectomy. Yoshimura andNogami2' presented some evidence for the presenceof adrenocorticotrophic hormone and thyroidstimulating hormone in stellate cells of the rat,though Shirasawa et a!6 showed the independence ofthese hormone secreting cells from folliculo-stellatecells. Other postulated functions of the folliculo-

stellate cells have been listed and include a suppor-tive function and phagocytic scavenger activity.6We have shown intense staining for GFAP, S 100,

and NSE within normal pituitary follicular struc-tures in man. We have also confirmed the intenseimmunoreactivity of the walls of the pars intermediacysts, the so called Rathke's cysts. We have foundintense reactivity for GFAP, S100, and NSE withinthe cells lining these cysts, as have others in man8 22

and in rat.5 The multiple reactivity of these cellssuggests that they may be pluripotential. Thefolliculo-stellate cells in the anterior lobe may bederived from these cyst areas and transform onreaching their destination to GFAP or S100 positivecells or they may even produce both proteins.Whether there is any additional pluripotential func-tion is debatable. As yet there is little strong evi-dence to support the theory that folliculo-stellatecells may differentiate into hormone secreting cellsother than the work on rats of Ohtsuka et al.20 Theevidence against transformation into a hormone sec-retory cell is the failure to show any transitional cellwith immune reactivity for S100, GFAP, and hor-mones. On some occasions we have seen follicularstructures with a definite lumen, which have stainedpositively for S100 (Fig. 8) and, rarely, similar struc-tures with a predominance of GFAP positive cells. Itis possible that these represent the transitional cell,but we have not yet attempted to show by doublelabelling whether these cells are also secretory.Neither the GFAP positive cells nor the S100 posi-tive cells in these follicular structures contain anyobvious granules.

It is noteworthy that the folliculo-stellate cells areNSE negative and the likelihood is then that they

486



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

487Immunocytochemistry offolliculo-stellate ceUs of normal and neoplastic human pituitary gland*; have noted that GFAP staining in stromal areas ismore intense than in follicular areas, which suggests

W* # changes in the intermediate filaments of thefolliculo-stellate cells. No such differences havebeen noted in S100 reactivity, and it is possible thatthese reflect metabolic changes and cell functionappropriate to their site, the S100 protein serving asa source for a variety of follicular activities requiringa carrier protein function. These changes are par-ticularly prominent in adenomas, where bothnumber and reactivity is greater than in the normalgland, and are similar to the changes described afterhormone manipulation in the rat.6

Fig. 8 Pituitary adenoma. Glial fibrillary acidic proteinpositive folliculo-stellate cells forming follicular structures.xSOO.

are of glial type and neuroectodermal in origin.Their negative staining for NSE indicates that theyare unlikely to be part of the APUD system as NSEis regarded as a specific marker for neurones andcells of the APUD system.'4 16 Our failure to showNSE in folliculo-stellate cells using anti-'yy enolaseantiserum supports the concept that these are gliallike cells which contain only the aa-isoenzyme.2324Our limited experience with the use of this tech-

nique on cultures suggests that this hypothesis is cor-rect since we have been able to show within a pitu-itary tumour culture a population of cells which are

strongly positive to GFAP but only weakly stainingto S100. These differences may be due to a matura-tion process and differential expression of S100.We have attempted to explore further the possible

neuroectodermal origin of the folliculo-stellate cellusing the monoclonal antibody UJ13A (the gener-

ous gift of Dr JT Kemshead), which is a neuroecto-dermal marker.25 In two cultures derived from pitu-itary adenomas with a high proportion of folliculo-stellate cells we were able to label these cells, whichconfirms their neuroectodermal origin (unpublishedobservations).

In speculating on the reason for the differentstaining properties of the folliculo-stellate cell at dif-ferent sites and especially the striking intensity ofstaining within tumours we tentatively suggest an

explanation based on a maturation process and a

differential expression of both S100 and GFAP. We

We thank the Neuropathology Department of theMidland Centre for Neurosurgery and Neurology(head of department Dr M Salmon) for providingsections and material and Veronica Turner for typ-ing the manuscript. We also thank Leitz for the loanof a Laborlux microscope and MPS45 photoautomatcamera system. This work was supported by fundsfrom the West Midlands Regional Health Authorityand the Mitchell Trust Fund, University of Birming-ham.

References

'Kovacs K, Horvath E, Ryan N. Immunocytology of the humanpituitary. In: DeLellis RA, ed. Diagnostic immunohistochemis-try. New York: Masson, 1981:17-35.

2 Bergland RM, Torack RM. An ultrastructural study of follicularcells in the human anterior pituitary. Am J Pathol 1969;57:273-97.

3 Horvath E, Kovacs K, Penz G, Ezrin C. Origin, possible functionand fate of 'follicular cells" in the anterior lobe of the humanpituitary. An electron microscopic study. Am J Pathol 1974;77:199-212.

4 Fukuda T. Agranular stellate cells (so-called follicular cells) inhuman fetal and adult adenohypophysis and in pituitaryadenoma. Virchows Arch [Pathol Anat] 1973;359: 19-30.

Nakajima T, Yamaguchi H, Takahashi K. S100 protein in fol-liculostellate cells of the rat pituitary anterior lobe. Brain Res1980; 191:523-31.

6 Shirasawa N, Kihara H, Yamaguchi S, Yoshimura F. Pituitaryfolliculo-stellate cells immunostained with S100 protein anti-serum in postnatal, castrated and thyroidectomized rats. CellTissue Res 1983;231:235-49.

7 Vila-Porcile E. Le reseau des cellules folliculo-stellaires et lesfollicules de radenohypohyse du rat (pars distalis). [The net-work of the folliculo-stellate cells and the follicles of theadenohypophysis of the rat (pars distalis)]. Z Zellforsch1972; 129:328-69.

Velasco ME, Roessmann U, Gambetti P. The presence of glialfibrillary acidic protein in the human pituitary gland. JNeuropathol Exp Neurol 1982;41: 150-63.

Suess U, Pliska V. Identification of the pituicytes as astroglialcells by indirect immunofluorescence staining for the glialfibrillary acidic protein. Brain Res 1981;221:27-33.

Salm AK, Hatton GI, Nilaver G. Immunoreactive glial fibrillaryacidic protein in pituicytes of the rat neurohypophysis. BrainRes 1982;236:471-6.

"Cocchia D, Miani N. Immunocytochemical localization of thebrain- specific S-100 protein in the pituitary gland of adult rat.J Neurocytology 1980;9:771-82.

.9

f._t

x f'

i8...



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

4882 Kahn HJ, Marks A, Thom H, Baumal R. Role of antibody to

S100 protein in diagnostic pathology. Am J Clin Pathol1983;79:341-7.

Odelstad L, Pahlman S, Lackgren G, Larsson E, Grotte G, Nils-son K. Neuron-specific enolase: a marker for differential diag-nosis of neuroblastoma and Wilm' s tumour. J Paed Surg1982; 17:381-85.

4 Dhillon AP, Rode J, Leatham A. Neurone specific enolase: anaid to the diagnosis of melanoma and neuroblastoma. His-topathology 1982;6:81-92.

'5 Bonnin JM, Rubinstein LJ. Immunohistochemistry of centralnervous system tumours. Its contributions to neurosurgicaldiagnosis. J Neurosurg 1984; 60: 1121-33.

Tapia FJ, Polak JM, Barbosa AJA, et al. Neuron-specific enolaseis produced by neuroendocrine tumours. Lancet 1981;i:808-11.

7Sternberger LA, Hardy PH, Cuculis JJ, Meyer HG. The unlabel-led antibody enzyme method of immunohistochemistry. Prep-aration and properties of soluble antigen-antibody complex(horseradish peroxidase-antihorseradish peroxidase) and itsuse in identification of spirochetes. J Histochem Cytochem1970; 18:315-33.

Sternberger LA. Immunocytochemistry. Englewood Cliffs, NewJersey: Prentice-Hall, 1974.

Yoshimura F, Soji T, Ato S, Yokoyama M. Development anddifferentiation of rat pituitary follicular cells under normal andsome experimental conditions with special reference to an

Morris, Hitchcock

interpretation of renewal cell system. Endocrinol Jpn1977;24:435-49.

20 Ohtsuka Y, Ishikawa H, Omoto T, Takasaki Y, Yoshimura Y.Effect of CRF on the morphological and functional differentia-tion of the cultured chromophobes isolated from rat anteriorpituitaries. Endocrinol Jpn 1971; 18: 133-53.

21 Yoshimura F, Nogami H. Immunohistochemical characterizationof pituitary stellate cells in rats. Endocinol Jpn 1980;27:43-51.

22 Asa SL, Ryan N, Kovacs K, Singer W, Marangos PJ. Immunohis-tochemical localization of neuron-specific enolase in thehuman hypophysis and pituitary adenomas. Arch Pathol LabMed 1984;108:40-3.

23 Schmechel D, Marangos PJ, Brightman M. Neurone-specificenolase is a molecular marker for peripheral and centralneuroendocrine cells. Nature 1978;276:834-6.

24 Royds JA, Parsons MA, Taylor CB, Timperley WR. Enolaseisoenzyme distribution in the human brain and its tumours. JPathol 1982; 137:37-49.

25 Allan PM, Garson JA, Harper EI, et al. Biological characteriza-tion and clinical applications of a monoclonal antibody recog-nizing an antigen restricted to neuroectodermal tissues. Int JCancer 1983;31:591-8.

Requests for reprints to: Professor ER Hitchcock, MidlandCentre for Neurosurgery and Neurology, Holly Lane,Smethwick, Warley, West Midlands B67 7JX, England.



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

Documents

Immunocytochemistry of cells of normal human · Glial fibrillary acidic protein and S100have been shownin the ... Of the six recognisable cell types within the human pituitary gland,