Evidence for Alternative Pathways of Granulosa CellDeath in Healthy and Slightly Atretic BovineAntral Follicles*

I. L. VAN WEZEL, A. M. DHARMARAJAN, T. C. LAVRANOS, AND R. J. RODGERS

Department of Medicine, Flinders University of South Australia (I.L.V.W., T.C.L., R.J.R.), SouthAustralia 5042; and the Department of Anatomy and Human Biology, University of Western Australia(A.M.D.), Nedlands, Western Australia 6907, Australia

ABSTRACTGranulosa cell death is an early feature of atresia; however, there

are many apparent contradictions in the literature concerning themode of granulosa cell death. We have therefore examined this pro-cess in bovine healthy and atretic antral follicles, using a variety ofestablished techniques. Light and electron microscopic observationsindicated the presence of pyknotic or shrunken nuclei in both themembrana granulosa and the antrum. In the membrana granulosa,these nuclei were frequently crescent shaped and uniformly electrondense and were approximately the same size as healthy nuclei, all ofwhich are typical of early apoptosis. However, these nuclei werewithin the membranes of a healthy granulosa cell, suggesting thatphagocytosis by a neighboring granulosa cell is an unusually earlyevent in the apoptotic pathway of granulosa cells. In the membranagranulosa, pyknotic nuclei stained intensely with hematoxylin butweakly with the DNA-intercalating stain propidium iodide. A per-

centage of these pyknotic nuclei stained by TUNEL (terminal deoxy-UTP nick end-labeling). However, in the antrum, the pyknotic nucleiand larger globules of DNA stained intensely with both hematoxylinand propidium iodide, but were not TUNEL positive. The comet assayof cell death produced a streak tail of randomly nicked DNA, ratherthan the plume of low mol wt apoptotic DNA. Globules collected fromfresh follicular fluid stained intensely with propidium iodide and wereshown by PAGE to contain DNA, the majority of which was high molwt. In conclusion, granulosa cells within the membrana granulosa dieby apoptosis, with phagocytosis by a neighboring cell preceding anypotential budding of the nucleus or cell itself. Granulosa cells near theantrum are sloughed off into the antrum, and their death has featuresmore consistent with that of other cell types that undergo death as aresult of terminal differentiation. (Endocrinology 140: 2602–2612,1999)

ATRESIA of ovarian follicles is an important process, asit accounts for the loss of over 99% of oocytes in the

mammalian ovary (1, 2). It is first recognized by the presenceof pyknotic nuclei in the membrana granulosa or antrum,indicative of cell death (3). In more advanced stages of atre-sia, the granulosa cells have degenerated, and the follicularfluid is filled with cellular debris (3). The mechanisms bywhich granulosa cells die in normal healthy follicles andduring atresia are currently under intense investigation, par-ticularly with the recent discoveries regarding mechanismsof cell death.

Cells generally die by one of three mechanisms: apoptosis,necrosis, or terminal differentiation leading to cell death, asoccurs in the skin during keratinization or in erythropoiesis.In apoptosis, the nucleus typically condenses and formsbuds, frequently with a crescent shape of condensed chro-matin. These, or the fragments of cells containing them, areoften referred to as apoptotic bodies and are the morpho-logical hallmark of apoptosis. The apoptotic bodies are eitherphagocytosed by macrophages or, in epithelia, phagocytosed

by neighboring cells or extruded into a lumen (4). In manyapoptotic cells, the DNA is degraded by specific endonucle-ases, producing DNA fragments that are multiples of nu-cleosome size (180 bp) in length. This partially degradedDNA can be detected as a ladder by gel polyacrylamideelectrophoresis (PAGE). It can also be detected in tissue sec-tions by end labeling the multitude of DNA ends by TUNEL(terminal deoxy-UTP nick end labeling) or related methods(5–7). During necrosis, the nucleus shrinks initially, but incontrast to apoptosis, it does not bud into apoptotic bodiesnor is its DNA degraded into multiples of nucleosomelengths. Instead, in the first instance the DNA is randomlynicked, which by gel electrophoresis produces a smear ofrandom sizes of DNA. Necrosis often involves more than onecell, and the cellular debris is removed by phagocytic mac-rophages (4). Terminal differentiation is an alternative mech-anism that in selected cell types, such as red blood cells (8)and the outer squamous layers of skin (9), involves expulsionor destruction of the nucleus before the cessation of cellularfunction leads ultimately to cell death. The question of whichof these three forms of cell death is involved in the death ofgranulosa cells in healthy and atretic follicles has yet to beanswered.

Much of the current literature emphasizes the a role ofapoptosis in granulosa cell death during atresia (10, 11).Thus, a ladder-like pattern typical of apoptotic cells wasproduced by gel electrophoresis using DNA from granulosacells (10, 12), and TUNEL and related techniques have la-

Received April 1, 1998.Address all correspondence and requests for reprints to: Dr. Ray-

mond J. Rodgers, Department of Medicine, Flinders University of SouthAustralia, Bedford Park, South Australia 5042, Australia. E-mail:ray.rodgers@flinders.edu.au.

* This work was supported by grants from the Flinders MedicalCenter Research Foundation, Flinders University of South Australia, theNational Health and Medical Research Council of Australia, and theRaine Foundation.

0013-7227/99/$03.00/0 Vol. 140, No. 6Endocrinology Printed in U.S.A.Copyright © 1999 by The Endocrine Society

2602

beled granulosa cells in many species [human (13, 14), cow(15), sheep (16), pig (17), rat (18), and rabbit (19)], all of whichsuggests that granulosa cells undergo apoptosis. However,little attempt has been made to determine where apoptosisoccurs in the membrana granulosa, and there are a numberof ambiguities and inconsistencies in this literature. Oppos-ing the idea that granulosa cells die by apoptosis, the gran-ulosa cells from neither dominant nor nondominant folliclesin one study of human ovaries were labeled by TUNEL (20),and ultrastructural studies have reported granulosa cellswith necrotic rather than apoptotic features (21, 22). Inex-plicably, in one study, only a portion of all of the pyknoticnuclei in the membrana granulosa of atretic follicles werelabeled by TUNEL (23). Furthermore, although pyknotic nu-clei have been observed in the membrana granulosa and theantrum of ovarian follicles using light microscopy (10, 24, 25),there have been no studies directly determining whetherthese are the result of apoptosis or necrosis.

Therefore, in the current study of healthy and atretic bo-vine antral follicles we have sought to examine the modes ofgranulosa cell death. We initially addressed this issue byexamining the ultrastructure of granulosa cells in healthyand slightly atretic follicles. To further examine the integrityof DNA in dying granulosa cells of healthy and atretic fol-licles, we then used a range of techniques, including pro-pidium iodide staining, TUNEL, an adaptation of the cometassay of cell death (26, 27), and PAGE. On the basis of all ofour observations, we propose that granulosa cell death oc-curs by more than one pathway, dependent upon the locationwithin the membrana granulosa.

Materials and MethodsTissues

All bovine ovaries or reproductive tracts used in this study werecollected at a local abattoir, within 20 min of slaughter, from cowsassessed visually as being nonpregnant unless stated otherwise. Oneovary from each of three reproductive tracts was fixed by perfusion andused for electron microscopy (see below), as previously described indetail (28). Additionally, half of each of six ovaries was fixed by im-mersion in Carnoy’s solution (60% ethanol, 30% chloroform, and 10%acetic acid; 1.5 h), washed overnight in 95% ethanol, and embedded inparaffin. Sections (5 mm) were deparaffinized in xylene, rehydrated indecreasing concentrations of ethanol, and stained with hematoxylin orboth hematoxylin and propidium iodide (below) for histological studyor used in the TUNEL assay (below). An additional two half-ovarieswere snap-frozen in Tissue-Tek OCT compound (Miles, Inc., Elkhart,IN). Sections (10 mm) were cut on a CM1800 Leica cryostat (Leica Corp.,Rockleigh, NJ) and used in the comet assay of cell death (below). Ad-ditionally, seven fresh ovaries from nonpregnant cows and five frompregnant cows were used for the collection of globules, as describedbelow.

Classification of follicular health

One section from each of the Carnoy’s-fixed, paraffin-embedded ova-ries was stained with hematoxylin, and each antral follicle that had amembrana granulosa (as opposed to extremely atretic follicles in whichthe membrana granulosa was absent) was examined. Pyknotic nuclei(round or crescent shaped) were identified according to standard de-scriptions and stained intensely with hematoxylin. The health of antralfollicles was initially determined according to the integrity of the mem-brana granulosa and the general quantity of pyknotic nuclei in this layer.Thus, healthy follicles were those with an intact membrana granulosaand few if any pyknotic nuclei in this layer; atretic follicles had a tattered-looking membrana granulosa and a moderate number of pyknotic nu-

clei; very atretic follicles had numerous pyknotic nuclei, and very fewnuclei in the membrana granulosa appeared healthy.

Electron microscopy

The method of processing tissue for electron microscopic examinationhas been described in detail previously (28). Briefly, one ovarian arteryper reproductive tract was cannulated, and blood was flushed from theovary with Earle’s balanced-salt solution. The ovary was then perfusedwith 2.5% glutaraldehyde in 0.1 m MOPS (pH 7.3). Incisions were sub-sequently made in regions close to antral follicles to allow greater pen-etration of the fixative, and the ovary was then placed in 2.5% glutar-aldehyde in 0.1 m MOPS (4-morpholino-propanesulfonic acid) buffer(pH 7.3) for a minimum of 24 h. Small wedges of tissue containing antralfluid as well as granulosa and thecal layers were cut from the folliclesand further processed, involving postfixation in osmium tetroxide, de-hydration in acetone, and embedding in epoxy resin. Thick sections werestained with methylene blue in sodium borate. Thin sections werestained with uranyl acetate and lead citrate and observed and photo-graphed using a JEOL CS 1200 electron microscope (Peabody, MA).

Nuclear staining observed by light microscopy

Tissue sections from paraffin-embedded ovaries that had beenstained with hematoxylin were counterstained with propidium iodide,which intercalates with DNA. This involved pretreatment of deparaf-finized tissue sections with 0.1% Triton X-100 (catalogue no. T-6878,Sigma Chemical Co., St. Louis, MO; 7 min), followed by rinses in PBS(10 mm sodium/potassium phosphate in 0.137 m NaCl and 5 mm KClsolution, pH 7.3; three times, 5 min each time), then incubation with 50mg/ml propidium iodide (catalogue no. P-4170, Sigma Chemical Co.; 45min, room temperature) in PBS, and a further rinse in PBS (5 min) beforemounting in buffered glycerol (0.167 m Na2CO3 in 67% glycerol, pH 8.6).Propidium iodide staining was visualized with an Olympus Corp.Vanox AHBT3 fluorescence microscope, using the G filter. Sections thathad been stained with both hematoxylin and propidium iodide were insome cases observed using the U filter on the same microscope, whichenabled visualization of both brightfield (i.e. hematoxylin staining) andfluorescence (i.e. propidium iodide) concurrently. Nuclei that stainedwith both hematoxylin and propidium iodide appeared white, whereasthose stained with hematoxylin only or stained only faintly with pro-pidium iodide appeared blue.

One randomly chosen follicle from each of five sections (one sectionper ovary) that had been stained with hematoxylin and propidiumiodide was examined in detail. Pyknotic nuclei in each follicle wereidentified under brightfield conditions, as nuclei with intense hema-toxylin staining. Each of these nuclei was then examined under theconditions showing both hematoxylin and propidium iodide staining(described above), and a count was made of the number of pyknoticnuclei that appeared white, indicating essentially intact DNA, and thosethat appeared blue, indicating degraded DNA.

TUNEL

Deparaffinized sections from each of the six ovaries fixed in Carnoy’ssolution were incubated in 3% H2O2 in methanol (30 min, room tem-perature) to block endogenous peroxidase activity. All sections weresubsequently rinsed in PBS (three times, 5 min each time), then incu-bated in 5 mg/ml proteinase K (catalogue no. 745723, Boehringer Mann-heim, Mannheim, Germany) in PBS (45 min, 37 C) and rinsed again inPBS (three times, 5 min each time). Sections were subsequently incu-bated with 0.5 nm digoxygenin-11–29-deoxy-uridine-59-triphosphate(dig-11-dUTP; catalogue no. 1093088, Boehringer Mannheim), 50 U/mlterminal transferase (catalogue no. 220582, Boehringer Mannheim), 1.5mm CoCl2 in buffer (30 mm Tris-Cl, pH 7.2, and 140 mm Na-cacodylate;1 h, 37 C). The dig-11-dUTP was omitted from negative control sections.Sections were rinsed in PBS (three times, 5 min each time), then incu-bated with a 1:100 dilution of mouse monoclonal anti-digoxygenin (cat-alogue no. 1333062, Boehringer Mannheim) in PBS (overnight, roomtemperature). After rinsing again in PBS (three times, 5 min each time),sections were incubated with 1:100 biotinylated horse antimouse IgG(catalogue no. BA-2000, Vector Laboratories, Inc., Burlingame, CA; 3 h,room temperature), rinsed in PBS (three times, 5 min each time), incu-

GRANULOSA CELL DEATH IN BOVINE FOLLICLES 2603

bated with 1:100 avidin-biotin complex from the Vectastain ABC kit(catalogue no. PK-4001, Vector Laboratories, Inc.) in PBS, rinsed in PBS(three times, 5 min each time), incubated with 0.7 mg/ml diaminoben-zine hydrotetrachloride (10 min, room temperature), then with 0.7mg/ml diaminobenzine hydrotetrachloride in 0.06 mg/ml urea H2O2 in0.06 m Tris buffer (SigmaFast, D-4418, Sigma Chemical Co.; 10 min, roomtemperature), and finally rinsed in PBS (twice, 5 min each time) andmounted with buffered glycerol. After observation and in some casesphotography, coverslips were floated off in PBS, and sections werecounterstained with hematoxylin.

Comet assay of cell death

Sections (10 mm) were cut from frozen ovaries and collected on fullyfrosted slides (Fisher Scientific, Pittsburgh, PA) that had been coatedwith a layer of 0.8% agarose (catalogue no. 200-0010, Progen Industries,Inc., Darra, Australia) in PBS, and a second layer of agarose was pre-pared over the section. Slides were then placed in cell lysis solution (0.25m NaCl, 0.01 m EDTA, 1 mm Tris, 0.1% sodium lauryl sarcosinate, and0.1% Triton X-100, pH 10.0) for 3 h at 4 C. They were subsequentlyimmersed in either alkaline electrophoresis buffer (0.1% 8-hydroxyquin-olone, 0.01 m EDTA, 0.30 m NaOH, and 0.02% dimethylsulfoxide, pH10.0) or nondenaturing electrophoresis buffer (0.1 m Tris buffer, pH 8.0,containing 0.09 m sodium borate and 1 mm EDTA) for 20 min at roomtemperature, electrophoresed in fresh electrophoresis buffer for 15 min(24 V), then washed in a solution of 0.4 m Tris (pH 7.5). Sections werefixed in ethanol (20 min), then stained with propidium iodide (50 mg/mlin PBS; 45 min), washed in PBS, and mounted with buffered glycerol forobservation by fluorescence microscopy.

Ceramide-treated 293 cells (American Type Culture Collection, Ma-nassas, VA) were used as an apoptotic control. These cells were culturedin DMEM-Ham’s F-12 medium (catalogue no. 50–327-PA, Trace Bio-sciences, Castle Hill, Australia) with 10% FCS and antibiotics (100 mg/mlpenicillin, 100 mg/ml streptomycin, and 0.25 mg/ml fungizone; CSL,Melbourne, Australia) in 5% CO2 in air, with or without addition of theapoptosis-inducing agent C6-ceramide (N-hexanoyl-d-sphingosine; cat-alogue no. H-6524, Sigma Chemical Co.) to the culture medium (10 mm,12 h). At the end of the culture period, cells were centrifuged to form apellet and were snap-frozen in OCT for use in the comet assay of celldeath, as described above.

Collection and identification of globules from antral fluid

In preliminary observations using an Olympus Corp. CK2 invertedmicroscope, it was observed that fluid that had been aspirated fromantral follicles frequently contained spherical globules, which were oftenseveral times larger than individual cells and apparently had no outerlimiting membranes. Therefore, the fluid was aspirated from all folliclesof each of three ovaries from nonpregnant cows and three ovaries frompregnant cows, using an 18-gauge needle and sterile 10-ml syringe. Foreach ovary, the fluid of all of its follicles was expelled into a 35-mm petridish and examined by inverted microscope and grid eyepiece. Thenumber and size of globules in the region of the grid were counted in10 random areas of each dish, and the number of globules per ml fluidwas calculated. The size of the globules was also recorded.

The fluid from an additional three ovaries was pooled, and globuleswere transferred by mouth pipette onto glass slides coated with poly-l-ornithine (Sigma Chemical Co.) and Fro-Tissuer (Probing and Struc-ture, Thuringowa Central, Australia) and air-dried overnight. The slideswere subsequently immersed first in Carnoy’s solution (30 min) and thenin absolute ethanol (5 min), and brief rinses in decreasing concentrationsof ethanol (95% and 70%) were performed before a final rinse in PBS (5min). They were then incubated with propidium iodide (50 mg/ml inPBS, 45 min) and examined by fluorescence microscopy.

To confirm that globules were composed of DNA and to determinethe integrity of this DNA, we collected the globules from an additionalsix ovaries (four from pregnant, two from nonpregnant cows) and ex-amined the DNA by PAGE. The globules were transferred by mouthpipette into microfuge tubes. As a small number of other ovarian cellswere inadvertently transferred at the same time, equivalent quantitiesof ovarian cells were collected into separate microfuge tubes as controlsamples. Cells were lysed, and protein was digested by incubation in thepresence of 200 mg/ml proteinase K (Boehringer Mannheim) in NET

buffer (200 mm NaCl, 20 mm Tris-HCl, and 2 mm EDTA, pH 7.4) for 2 hat 37 C. The cell lysate was extracted with phenol-chloroform (29). TheDNA was precipitated with ethanol (2–3 vol) with (20 mg, 220 C over-night) or without (220 C, 1 h) the addition of carrier transfer RNA. Thefinal DNA pellet was dried under vacuum (15 min, Speed-Vac, SavantInstrument Co., Farmingdale, NY) and dissolved in 20 ml sterile water.Aliquots were mixed with loading buffer and electrophoresed on a 6%polyacrylamide gel in TBE buffer (0.09 m Tris, 0.09 m borate, and 2 mmEDTA) (29). DNA fragment sizes were estimated using a 100-bp ladder(Pharmacia Biotech). After electrophoresis, the gel was stained withethidium bromide (1 mg/ml) and destained in water, then visualizedunder UV and photographed using Polaroid film under transillumi-nation.

Photography

All photographs were taken using Kodak T-Max 400 black and whitefilm (Eastman Kodak Co., Rochester, NY). An Olympus C35AD-4 cam-era attachment was used with the Olympus Vanox AHBT3 fluorescencemicroscope, and an Olympus SC35 camera attachment was used withthe Olympus BX50 microscope and the Olympus CK2 inverted micro-scope (Olympus Corp., New Hyde Park, NY).

ResultsGeneral

It was observed that the membrana granulosa was struc-tured. Cells in different regions of the membrana granulosawere of different shapes, and cell shape corresponded withcell location within the membrana granulosa; columnar cellswere frequently observed in the one or two layers lining thebasal lamina, rounded cells were located in the middle layersof the membrana granulosa, and flattened granulosa cellswere frequently observed in the layer closest to the antrum(Fig. 1). From our studies it is clear that healthy or atreticantral follicles contained degraded nuclei of two types. Thosemore prevalent in the middle layers of the membrana gran-ulosa were pyknotic. Near the antral surface of the mem-brana granulosa and in the antral follicular fluid there wereboth pyknotic nuclei and large globular DNA-containingstructures. Each of these types is described separately below.

Pyknotic nuclei in the middle layers of themembrana granulosa

Within the membrana granulosa, pyknotic nuclei werepredominantly in the central region of the membrana gran-ulosa, rather than in the basal layer or the layer closest to theantrum. Approximately one third of these pyknotic nucleiwere crescent shaped (a shape typical of apoptotic nuclei),and with few exceptions the remainder were spherical anduniformly dense (not necessarily apoptotic nuclei).

Electron-dense structures in the membrana granulosa, in-terpreted as the pyknotic nuclei evident by light microscopy,were clearly visualized by electron microscopy (Figs. 1 and2). The usual form of these structures was a crescent orrounded shape, which stained homogeneously, being moreelectron dense than the nuclei of adjacent healthy granulosacells, but accompanied by small clumps of more electron-dense granular material (Fig. 2, a–c). In most cases, thesepyknotic nuclei were located within the cytoplasm of a cellthat had its own apparently healthy nucleus and intact or-ganelles including mitochondria, endoplasmic reticulum,and lipid (Fig. 2, a–c). This description fits the idea that onecell that had (or went on to develop) a pyknotic nucleus was

2604 GRANULOSA CELL DEATH IN BOVINE FOLLICLES Endo • 1999Vol 140 • No 6

phagocytosed by a healthy granulosa cell. The pyknotic nu-clei were approximately the same diameter as the healthynuclei, and often retained some semblance of a nuclear mem-brane. Surrounding the nuclear membrane were numerousclosely packed organelles that resembled mitochondria in

size; some of these had cristae. Endoplasmic reticulum wasalso evident in this region. The pyknotic nucleus and closelypacked organelles were all separated from the cytoplasm ofthe healthy cell by a membrane system, probably the remainsof the original cell membrane.

In cases where the pyknotic nucleus was not within thebounds of a cell that had its own healthy nucleus, the cellularorganelles had integrity and a spatial orientation equivalentto those of healthy granulosa cells. That is, they were easilyidentified as mitochondria, endoplasmic reticulum, or lipidand were not tightly packed together around the nucleus.The size of the cell was equivalent to that of adjacent healthycells.

Pyknotic nuclei in the antral layers of the membranagranulosa and in the antrum

Using light microscopy, structures that stained with he-matoxylin and resembled pyknotic nuclei were also presentin the antrum, often loosely associated with the antral layerof granulosa cells (see Fig. 3, a and c, and Fig. 4, c and e).These were usually spherical and ranged in size up to ap-proximately 5 times the diameter of a normal granulosa cellnucleus. However, the largest of these frequently containedvacuoles that did not stain with hematoxylin. Crescent-shaped pyknotic nuclei were occasionally seen in associationwith the membrana granulosa, but rarely in the antrumproper.

Electron microscopic examination of the antrum near themembrana granulosa revealed a number of spherical bodiesranging in size from approximately 2–4 mm in diameter.These were of three types. Most commonly, they consisted ofa very electron-dense sphere, surrounded by a large numberof closely packed organelles, including mitochondria-likeorganelles, endoplasmic reticulum, and lipid, all enclosedwithin a membrane (Fig. 2d). Occasionally, electron-densespheres resembling those just described, but without thesurrounding organelles and membranous envelope, wereobserved (Figs. 1 and 2d). It is likely that this second typeresulted from destruction of the first type during tissue pro-cessing, as membranous matter and organelles were seen inclose association with some type 2 structures (Fig. 2d).Thirdly, structures were seen that contained no electron-dense sphere in the plane of section, but were composed ofswirled, branching membranes (Fig. 2e).

Although granulosa cells of preantral follicles and antralfollicles that were very atretic did display ultrastructuralfeatures of necrosis (not shown), in the current study ofhealthy or slightly atretic antral follicles, granulosa cells dis-playing lysed organelles or clumped chromatin were notseen in the membrana granulosa.

Light microscopy

The six Carnoy’s solution-fixed ovaries contained a total of17 healthy antral follicles and 8 atretic antral follicles withremnants of a membrana granulosa. Additionally, there were12 follicles in later stages of atresia that did not containgranulosa cells and were therefore not further considered inthis study.

FIG. 1. Electron micrograph of the membrana granulosa of a bovineantral follicle, showing columnar granulosa cells lining the basallamina, more rounded granulosa cells closer to the antrum, and flat-tened granulosa cells adjacent to the antrum (A). Pyknotic nuclei arepresent both within the membrana granulosa and in the follicularantrum (small arrow). The large arrow indicates the follicular basallamina. Bar, 5 mm.

GRANULOSA CELL DEATH IN BOVINE FOLLICLES 2605

Dual staining with hematoxylin and propidium iodideDual staining with hematoxylin and propidium iodide

was carried out to determine whether all of the structuresthat stained intensely with hematoxylin were indeed rich in

DNA and thus derived from nuclei (Fig. 3). All nuclei ofhealthy granulosa cells that stained with hematoxylin alsostained with propidium iodide, including mitotic nuclei.Pyknotic nuclei in the membrana granulosa stained with

FIG. 2. Electron micrographs of bovine antral follicles, showing different forms of granulosa cell death. a–c, Within the membrana granulosa,nuclei with features typical of apoptosis (arrowheads) within the cell membranes (arrows) of granulosa cells containing their own apparentlyhealthy nuclei (N); d, pyknotic nuclei (arrowhead) within atretic bodies present in the follicular antrum (A), adjacent to the membrana granulosa;e, membranous atretic body present in the follicular antrum. Bars, 1 mm.

2606 GRANULOSA CELL DEATH IN BOVINE FOLLICLES Endo • 1999Vol 140 • No 6

propidium iodide, but this staining was usually less intensethan staining of nuclei from healthy cells, in contrast to thehematoxylin staining of pyknotic nuclei, which often ap-peared more intense in pyknotic than in healthy nuclei (Fig.3). When the filters in the fluorescence microscope wereadjusted to enable visualization of both hematoxylin stainingand propidium iodide concurrently (Fig. 3d), 18% of thepyknotic nuclei in the membrana granulosa appeared blue

(hematoxylin staining stronger than propidium iodide stain-ing) rather than white (both). In most cases, return to thefluorescence filter alone revealed that these nuclei did stainwith propidium iodide, but very weakly. In only a few cases,was no propidium iodide staining evident.

In contrast to the pyknotic nuclei in the membrana gran-ulosa, the nuclei in the antrum and those loosely associatedwith the membrana granulosa stained very intensely with

FIG. 3. Dual staining of antral follicleswith hematoxylin (originally blue) andpropidium iodide (originally red fluo-rescent). In each panel, the membranagranulosa lies between the follicularbasal lamina (arrow) and the antrum(A). a and b, One section from a healthy/slightly atretic follicle; c–e, one sectionfrom an atretic follicle. Filters were ad-justed to show hematoxylin stainingonly (a and c), both hematoxylin andpropidium iodide staining (d), or pro-pidium iodide staining only (b and e).Pyknotic nuclei within the membranagranulosa stain intensely for hematox-ylin, but faintly for propidium iodide(small arrowheads), whereas globulesin the antrum stain intensely with bothstains (large arrowheads). Bar, 50 mm.

GRANULOSA CELL DEATH IN BOVINE FOLLICLES 2607

propidium iodide (Fig. 3, b and e) and with the same patternas that of staining by hematoxylin.

TUNEL in the membrana granulosa

Punctate brown staining was observed in the membranagranulosa of follicles in each of the treated sections, but wasabsent from negative-control parallel sections in which thedig-11-dUTP was omitted from the TUNEL protocol (Fig. 4).Little if any staining was observed in healthy follicles (Fig. 4a)compared with atretic follicles (Fig. 4b). Stained granulosacells were located throughout the membrana granulosa ofatretic follicles, but predominantly in the central region, con-sistent with our observations of the location of pyknoticnuclei. However, the pyknotic nuclei loosely associated with

the membrana granulosa or more freely distributed in thefollicular antrum rarely stained, except at the periphery ofvacuole-like structures in the largest of these structures (Fig.4d). Counterstaining with hematoxylin revealed that not allof the structures identified as pyknotic nuclei by hematoxylinstaining were TUNEL positive (Fig. 4, compare b with c andd with e).

The comet assay of cell death

293 cells were used as control cells. In either alkaline de-naturing buffer or in nondenaturing buffer, untreated cellsproduced no tails (not shown); cells treated with ceramideproduced the full plume tails of apoptosis (Fig. 5a). Antralfollicles in the bovine ovaries were unable to be sectionedat less than 10 mm due to the fragility of the frozen fol-licular fluid. The nuclei of granulosa cells are located veryclose to each other, and when 10-mm sections were usedin the assay of cell death, it was impossible to distinguishbetween the nuclei and the tails. However, the globules inthe antrum were more dispersed than the nuclei in themembrana granulosa, and their tails were streak-like (Fig.5b) even in alkaline denaturing buffer. This indicates thatthe DNA was of high mol wt, with very little nicked DNA,and that DNA was not degraded by the process ofapoptosis.

Globule identity

Globules present in freshly aspirated follicular fluidwere transparent and colorless and ranged from being

FIG. 4. Staining of pyknotic nuclei in antral follicles by TUNEL (originally dark brown). a, Healthy antral follicle with no labeled cells(counterstaining with hematoxylin revealed no pyknotic nuclei; not shown). b–e, Atretic follicles showing staining by TUNEL (b and d) or byTUNEL and then hematoxylin (c and e). Large arrowheads indicate pyknotic nuclei evident by hematoxylin that stain with TUNEL, whereassmall arrowheads indicate pyknotic nuclei evident by hematoxylin that do not stain with TUNEL. f, Negative control of an atretic antral follicle;digoxygenin-11-dUTP was omitted. Bar, 50 mm.

FIG. 5. Staining of DNA using the comet assay of cell death. In a andb, essentially undegraded DNA retains its own nucleus-like config-uration (arrowheads). In a, 293 cells that had been treated withceramide to induce apoptosis exhibit a plume tail (arrow), whereas inb, globules in the antrum of bovine follicles produce a thin smearemanating from the nucleus (arrow).

2608 GRANULOSA CELL DEATH IN BOVINE FOLLICLES Endo • 1999Vol 140 • No 6

completely smooth and uniform in appearance to havingone or more vacuoles (Fig. 6a) and appearing bubbly.When the globules were air-dried onto glass slides, thevacuoles were still evident (Fig. 6b). The globules rangedin size, with the largest observed being 620 mm. Theirmean size was 40 6 8 mm (mean 6 sem). The number ofglobules was estimated at 1200/ml follicular fluid. Theappearance of the globules did not differ between non-pregnant and pregnant animals.

Globules and ovarian cells that were collected onto glassslides and fixed with Carnoy’s solution stained positively

with propidium iodide (Fig. 6, c and d). The globules stainedmuch more intensely than did oocytes; zona pellucidae werenot stained by propidium iodide.

On PAGE (Fig. 6e), the DNA globules from ovaries ofnonpregnant or pregnant cows appeared as large smearsindicative of necrotic DNA, with only a faint ladder-likepattern indicative of apoptotic DNA. Thus, the vast majorityof the DNA in these globules, derived purely from deadgranulosa cells, was not degraded by the process of apopto-sis. The control samples, containing small numbers of ovar-ian cells, produced no bands (Fig. 6a). We estimate from the

FIG. 6. Identity of globules. Globulepresent in freshly aspirated fluid (a),after air-drying onto a glass slide (b;between arrowheads, vesicle indicatedwith an arrow; photographed with No-marski differential interference optics),and stained with propidium iodide (cand d; arrowheads). Note the lack ofstaining of the zona pellucida (arrows)of a naked oocyte. Bars 5 50 mm in a,100 mm in b–d. e, PAGE of recoveredglobules in lanes 1, 2, and 4. Lanes 3 and5 are negative controls (lane 3 is me-dium only; lane 5 contains a small num-ber of granulosa cells). The location ofthe transfer RNA used as a carrier inthe preparation step is indicated. M is aDNA ladder in steps of 100 bp.

GRANULOSA CELL DEATH IN BOVINE FOLLICLES 2609

approximate yields of DNA that the globules are aggregatesderived from large numbers (up to thousands) of dead gran-ulosa cells.

Discussion

The results reported in this manuscript support the hy-pothesis that in healthy and early atretic follicles, granulosacells within the middle layers of the membrana granulosa dieby a pathway different from that of granulosa cells close tothe antrum. Those within the middle layers undergo apo-ptosis, showing the typical crescent-shaped condensation ofchromatin, but apparently involving phagocytosis beforeany potential budding of the nucleus and cell itself. In con-trast, cells within the antum, which had presumably origi-nated from the granulosa layer(s) closest to the antrum, ex-hibited features more consistent with cell death resultingfrom terminal differentiation, as previously described forother cell types.

The study of apoptosis of granulosa cells has become avery popular topic in recent years (11). Surprisingly, though,none of the landmark papers in this area have considered theultrastructure of these cells in situ, given the importance ofmorphology for defining apoptosis (4). TUNEL and relatedtechniques have been used in previous studies by Billig et al.(30) and Chun et al. (31), and although these techniquespreferentially label apoptotic cells, they also label necroticcells (32, 33). Hence, they should not be used alone to dif-ferentiate between these two forms of cell death. The pro-duction of a ladder-like pattern by gel electrophoresis (10, 12,30, 31, 34, 35). Also, this method requires a pool of cells. Bymaking ultrastructural observations of condensed nuclei ofcrescent or circular shape, we now have very conclusiveevidence for a form of apoptosis in the membrana granulosaof bovine follicles. This was observed to occur most fre-quently in the middle layers of the membrana granulosa, theregion also observed to contain dividing cells (van Wezel,I. L., and R. J. Rodgers, unpublished observations).

In the classic process of apoptosis, nuclear condensationprecedes budding of the nucleus and of the cell itself intoseveral small apoptotic bodies (4). In epithelia (4), theseapoptotic bodies are rapidly phagocytosed by neighboringepithelial cells, and thus, free apoptotic bodies are rarelyseen. However, it is common to observe these small vesiclesonce they have been phagocytosed. Phagocytosis has nowbeen observed in ovine (36) and bovine follicles. The obser-vations we have made suggest that cells of the membranagranulosa of bovine follicles do not follow all of these classicprocesses precisely. Apoptotic nuclei found in the middlelayers of the membrana granulosa were similar in size tonormal nuclei, and thus had presumably not budded. How-ever, these were commonly seen within a granulosa cell thatcontained its own apparently healthy nucleus, indicatingthat the healthy cell had phagocytosed the apoptotic cell. Wethus hypothesize that either the apoptotic nuclei do not havethe capacity to bud or, more likely, that phagocytosis is anearly event in the apoptosis of bovine granulosa cells, pre-ceding any potential budding. Thus, the endonuclease sys-tem of the apoptotic nuclei, if activated, would potentially bedestroyed in the lysosomes of the phagocytosing cell. This

could explain why only a proportion of apoptotic nuclei waslabeled by TUNEL as observed here and previously (23).

During follicular atresia, the atrophy of granulosa cells hasbeen reported to involve features of necrosis (21, 22). Wehave observed in advanced stages of atresia of antral folliclesand in preantral follicles the presence of free organelles fromlysed cells and granulosa cells with lysed organelles andclumped chromatin typical of necrosis (van Wezel, I. L., andR. J. Rodgers, unpublished observations). The current studiesdo not rule out a role for necrosis in advanced stages offollicular atresia, perhaps as a widespread mechanism bywhich the membrana granulosa is ultimately destroyed.However in the present ultrastructural study of healthy fol-licles and follicles that were only slightly atretic, no necroticcells were observed in the membrana granulosa.

In the antrum we observed globules of DNA and pyknoticnuclei, frequently in membrane-bound bodies. Collectively,these structures have previously been called atretic bodies.They were originally described as arising from the swellingof cells at the surface of the membrana granulosa, followedby the dissolution of the cell membrane and later the nuclearmembrane (24). It is likely that the DNA globules, as we havecalled them, are derived from aggregates of variable num-bers of pyknotic nuclei after dissolution of the cell mem-brane. In the sheep, Hay et al. (36) reached a similar conclu-sion. However, the description of these atretic bodies (24) isfar from consistent with apoptosis, yet in one of the landmarkpapers investigating apoptosis by molecular means (10), theatretic bodies were renamed apoptotic bodies. Are theseatretic bodies really apoptotic? In support of the idea thatthey are, release of apoptotic bodies into a nearby lumen hasbeen observed in other organs (4), and in the current study,we observed that nuclei in this region were uniformly elec-tron dense, consistent with apoptosis, rather than exhibitingclumping of the nuclear chromatin typical of necrosis. How-ever, we obtained evidence that they do not undergo the highdegree of DNA fragmentation typical of apoptosis. Pro-pidium iodide staining of nuclei in the antrum was intense,and the globules or pyknotic nuclei in the antrum were notTUNEL positive, suggesting that the DNA was not exces-sively degraded, especially compared with that of pyknoticnuclei observed within the membrana granulosa. In thecomet assay of cell death, the globules in the antrum pro-duced a streak tail of randomly cut DNA rather than a plumeof apoptotic DNA, and globules of DNA that were presentin follicular fluid isolated from fresh ovaries were shown byPAGE to be mostly high mol wt rather than producing aladder-like pattern of apoptotic DNA. It is possible thatapoptosis of these cells had taken place without fragmenta-tion of the DNA into multiples of nucleosome length. How-ever, as no crescent-shaped nuclei were observed, and therewas no evidence of DNA fragmentation to the degree com-monly expected with apoptosis, more evidence is requiredbefore concluding that these atretic bodies are apoptoticbodies.

If the antral atretic bodies are not formed by apoptosis,then two alternative mechanisms for the formation ofpyknotic nuclei or atretic bodies remain: necrosis or terminaldifferentiation preceding cell death. The ultrastructure of theatretic bodies was not consistent with necrosis, as none of the

2610 GRANULOSA CELL DEATH IN BOVINE FOLLICLES Endo • 1999Vol 140 • No 6

usual features, such as dissolution of the cell membranes andlysis of organelles, was observed here. The remaining alter-native is terminal differentiation, comparable to the forma-tion of mature erythrocytes (8) or mature keratinocytes inskin (9). The changes in DNA during terminal differentiationthat ultimately result in cell death are not well understoodand may subsequently be shown to share some features incommon with apoptosis. The formation of the reticulocytesrequires the progressive condensation of nuclear material toform electron-dense pyknotic nuclei, similar to the granulosanuclei observed in the follicular antrum. These nuclei becomepolarized in the cell and are subsequently extruded, leavingthe mature reticulocyte (8). The extruded nucleus is subse-quently phagocytosed (37). We are not aware of any studiesexamining the nature of DNA degradation during terminaldifferentiation in other organs. However, in contrast to ap-optosis, terminal differentiation is associated with cellularloss of the ability to divide (9, 38), and in a related study weobserved that mitotic figures were less frequently observedin the antral portion of the membrana granulosa than in themiddle portions (van Wezel, I. L., and R. J. Rodgers, unpub-lished observations). Thus, the features of the atretic bodiesare highly consistent with terminal differentiation. The con-cept of terminal differentiation leading ultimately to celldeath has not previously been considered in the context ofovarian follicles.

Both the current results and other unpublished observa-tions support the concept that the membrana granulosa isdynamic and highly structured and has a number of featuresin common with the skin epidermis. This is illustrated in Fig.7. The basal/antral structure of some small antral follicles issimilar to that of skin; the granulosa cells closest to the basallamina are columnar, those cells slightly further from thebasal lamina are more rounded, and those cells furthest fromthe basal lamina are often flattened (24). This profile mightnot be static, as the columnar basal cells might becomerounded as the follicle enlarges. Like skin, cell division ispredominantly within a zone of the epithelium, but whereasthis zone is basal in skin, in the membrana granulosa it isfound in the middle region. Cells closer to the antrum are

older and become progressively flattened in shape, as occursin skin. The cells at the surface slough off, as occurs in skin,and this process leads to death of these cells. Our examina-tion suggests that the mode of death of these cells closest tothe antrum does not follow the process of apoptosis or ne-crosis but, rather, follows that of terminal differentiation.Skin also has a population of stem cells, and we have pos-tulated that granulosa cells arise from stem cells (38–40). Insupport of this, we have demonstrated that a small propor-tion of granulosa cells have at least one property of stem cells,namely the ability to divide without anchorage in vitro(38–40).

One major difference between skin epidermis and themembrana granulosa is that the latter can undergo wholesaledestruction, as occurs during follicular atresia. We observedthat cell death in early atretic follicles is localized to what wasthe replicative zone (van Wezel, I. L., and R. J. Rodgers,unpublished observations). Here, we have shown that thesecells died by apoptosis. Given that a cell makes a decision inG1 to either enter S or undergo apoptosis (41), it is likely thatan early process in atresia involves entry of the dividinggranulosa cell into apoptosis. As all other cells are predom-inantly in G0, especially those in nonreplicative zones, theirmode of death probably involves different biochemical pro-cesses; certainly, our observations here would support thisidea.

In conclusion, the pyknotic nuclei evident in hematoxylin-stained tissue sections of healthy and slightly atretic ovarianfollicles represent two distinct pathways of cell death. Thosein the middle layers of the membrana granulosa undergoapoptosis, but with phagocytosis apparently preceding anypotential budding, whereas atretic bodies in the antrum areapparently derived from the sloughing off of cells from thegranulosa layers closest to the antrum. These bodies sharefeatures in common with other cell types undergoing a ter-minal differentiation that results in cell death. Necrosis onlyappears to play a greater part in advanced atresia. Thesefindings explain many of the conflicting observations andconclusions made previously about follicular atresia. Ourfindings also raise the questions of why the membrana gran-

FIG. 7. Our current model for the structure and development of the membrana granulosa as a dynamic epithelium. a, Cells in different regionsof the membrana granulosa differ in shape, often being columnar in the basal zone, rounded in the middle layers, and flattened closest to theantrum. Those in the middle layers are proliferative and die by apoptosis. Closer to the antrum, the cells differentiate, slough off into the antrum,and die by terminal differentiation rather than apoptosis. The stem cells are proposed to be near the follicular basal lamina, in the most basalor suprabasal layers. b, With increasing follicle size, the overall structure of the membrana granulosa alters from many follicles having columnarcells in the basal zone, rounded cells in the middle zone, and often flattened granulosa cells in the antral zone to many follicles having roundedcells throughout the membrana granulosa.

GRANULOSA CELL DEATH IN BOVINE FOLLICLES 2611

ulosa is structured the way it is, and why it undergoes thechanges it does. These changes occur as the follicular fluidaccumulates in the expanding antrum, a process about whichwe know very little. It remains to be seen whether theseprocesses are all functionally related.

Acknowledgments

We sincerely thank Dr. Glenda Gobe (University of Queensland,Brisbane, Australia) for examining some of our electron micrographsand for many helpful discussions concerning the nature of apoptosis.Many thanks to Fiona Young, for providing us with her protocol for theTUNEL assay, and to Michael Bawden for assisting in DNA analyses. Wealso appreciate the generosity of Dr. Jeff Trahair (University of Adelaide,Adelaide, Australia) and Dr. Doug Brooks (Women’s and Children’sHospital, Adelaide, Australia), who donated reagents for work con-nected with this study.

References

1. Byskov AG 1978 Follicular atresia. In: Jones RE (ed) The Vertebrate Ovary.Plenum Press, New York, pp 533–562

2. Greenwald GS, Terranova PF 1988 Follicular selection and its control. In:Knobil E, Neill JD, Ewing LL, Greenwald GS, Markert CL, Pfaff DW (eds) ThePhysiology of Reproduction. Raven Press, New York, pp 387–445

3. Rajakoski E 1960 The ovarian follicular system in sexually mature heifers withspecial reference to seasonal, cyclical and left-right variations. Acta Endocrinol[Suppl 52] (Copenh) 34:6–68

4. Kerr JFR, Gobe GC, Winterford CM, Harmon BV 1995 Anatomical methodsof cell death. Methods Cell Biol 46:1–27

5. Gavrieli Y, Sherman Y, Ben-Sasson SA 1992 Identification of programmed celldeath in situ via specific labeling of nuclear DNA fragmentation. J Cell Biol119:493–501

6. Gold R, Schmied M, Rothe G, Zischler H, Breitschope H, Wekerle H, Lass-mann H 1993 Detection of DNA fragmentation in apoptosis: application of insitu nick translation to cell culture systems and tissue sections. J HistochemCytochem 41:1023–1030

7. Gold R, Schmied M, Giegerich G, Breitschopf H, Hartung HP, Toyka KV,Lassmann H 1994 Differentiation between cellular apoptosis and necrosis bythe combined use of in situ tailing and nick translation techniques. Lab Invest71:219–225

8. Castoldi GL, Beutler E 1988 Erythrocytes. In: Zucker-Franklin D, Greaves MF,Grossi CE, Marmont AM (eds) Atlas of Blood Cells: Function and Pathology,ed 2. Edi Ermes, Milan, pp 47–156

9. Stenn KS 1983 The skin. In: Weiss L (ed) Histology: Cell and Tissue Biology,ed 5. Elsevier, New York, pp 569–606

10. Hughes Jr FM, Gorospe WC 1991 Biochemical identification of apoptosis(programmed cell death) in granulosa cells: evidence for a potential mecha-nism underlying follicular atresia. Endocrinology 129:2415–2422

11. Tilly JL 1997 Apoptosis and the ovary: a fashionable trend or food for thought?Fertil Steril 67:226–228

12. Tilly JL, Kowalski KI, Johnson AL, Hsueh AJW 1991 Involvement of apo-ptosis in ovarian follicular atresia and postovulatory regression. Endocrinol-ogy 129:2799–2801

13. Yuan W, Giudice LC 1997 Programmed cell death in human ovary is a functionof follicle and corpus luteum status. J Clin Endocrinol Metab 82:3148–3155

14. Kugu K, Ratts VS, Piquette GN, Tilly KI, Tao XJ, Martimbeau S, AberdeenGW, Krajewski S, Reed JC, Pepe GJ, Albrecht ED, Tilly JL 1998 Analysis ofapoptosis and expression of bcl-2 gene family members in the human andbaboon ovary. Cell Death Differ 5:67–76

15. Jolly PD, Tisdall DJ, Heath DA, Lun S, McNatty KP 1994 Apoptosis in bovinegranulosa cells in relation to steroid synthesis, cyclic adenosine 3959-mono-phosphate response to follicle-stimulating hormone and luteinizing hormone,and follicular atresia. Biol Reprod 51:934–944

16. Murdoch WJ 1995 Programmed cell death in preovulatory ovine follicles. BiolReprod 53:8–12

17. Derecka K, Kalamarz H, Ziecik AJ 1995 Does apoptosis occur during follicularatresia in the follicle walls of the porcine ovary? Reprod Dom Anim 30:32–35

18. Palumbo A, Yeh J 1994 In situ localization of apoptosis in the rat duringfollicular atresia. Biol Reprod 51:888–895

19. Kasuya K 1995 The process of apoptosis in follicular epithelial cells in the rabbitovary, with special reference to involvement by macrophages. Arch HistolCytol 58:257–264

20. Funayama Y, Sasano H, Suzuki T, Tamura M, Fukaya T, Yajima A 1996 Cellturnover in normal cycling human ovary. J Clin Endocrinol Metab 81:828–834

21. Priedkalns J, Weber AF 1968 Ultrastructural studies of the bovine Graafianfollicle and corpus luteum. Z Zellforsch 91:554–573

22. Hubbard CJ, Oxbury B 1991 Follicular atresia. In: Familiari G, Makabe S, MottaPM (eds) Ultrastructure of the Ovary. Kluwer, Boston, pp 273–286

23. D’Herde K, De Pestel G, Roels F 1994 In situ end labelling of fragmented DNAin induced ovarian atresia. Biochem Cell Biol 72:573–579

24. Marion GB, Gier HT, Choudary JB 1968 Micromorphology of the bovineovarian follicular system. J Anim Sci 27:451–465

25. Koering MJ, Goodman AL, Williams RF, Hodgen GD 1982 Granulosa cellpyknosis in the dominant follicle of monkeys. Fertil Steril 37:837–844

26. Ostling O, Johanson KJ 1984 Microelectrophoretic study of radiation-inducedDNA damages in individual mammalian cells. Biochem Biophys Res Commun123:291–298

27. Fairbairn DW, Olive PL, O’Neill KL 1995 The comet assay: a comprehensivereview. Mutat Res 339:37–59

28. van Wezel IL, Rodgers RJ 1996 Morphological characterization of bovineprimordial follicles and their environment in vivo. Biol Reprod 55:1003–1011

29. Maniatis T, Fritsch EF, Sambrook J 1982 Molecular Cloning. Cold SpringHarbor Laboratory Press, Cold Spring Harbor

30. Billig H, Furuta I, Hsueh JW 1994 Gonadotropin-releasing hormone directlyinduces apoptotic cell death in the rat ovary: biochemical and in situ detectionof deoxyribonucleic acid fragmentation in granulosa cells. Endocrinology134:245–252

31. Chun S-Y, Eisenhauer KM, Minami S, Billig H, Perlas E, Hseuh AJW 1996Hormonal regulation of apoptosis in early antral follicles: follicle-stimulatinghormone as a major survival factor. Endocrinology 137:1447–1456

32. Ansari B, Coates P, Greenstein B, Hall P 1993 In situ end labelling detectsDNA strand breaks in apoptosis and other physiological and pathologicalstates. J Pathol 170:1–8

33. Grasl-Kraupp B, Ruttkay-Nedecky B, Koudelka H, Bukowska K, Bursch W,Schulte-Hermann R 1995 In situ detection of fragmented DNA fails to dis-criminate among apoptosis, necrosis and autolytic cell death: a cautionarynote. Hepatology 21:1465–1468

34. Oberhammer F, Fritsch G, Schmied M, Pavelka M, Printz D, Purchio D,Lassman H, Schulte-Hermann R 1993 Condensation of the chromatin at themembrane of an apoptotic nucleus is not associated with activation of anendonuclease. J Cell Sci 104:317–326

35. Falcieri E, Martelli AM, Bareggi R, Cataldi A, Cocco L 1993 The protein kinaseinhibitor staurosporine induces morphological changes typical of apoptosis inMolt-4 cells without concommitant DNA fragmentation. Biochem Biophys ResCommun 193:19–25

36. Hay MF, Cran DG, Moor RM 1976 Structural changes occurring during atresiain sheep ovarian follicles. Cell Tissue Res 169:515–529

37. Weiss L 1988 The life cycle of blood cells. In: Weiss L (ed) Cell and TissueBiology: A Textbook of Histology, ed 6. Urban and Schwarzenberg, Baltimore,pp 445–466

38. Lavranos TC, Rodgers HF, Bertoncello I, Rodgers RJ 1994 Anchorage-inde-pendent culture of bovine granulosa cells: the effects of basic fibroblast growthfactor and dibutyryl cAMP on cell division and differentiation. Exp Cell Res211:245–251

39. Rodgers RJ, Vella CA, Rodgers HF, Scott K, Lavranos TC 1996 Production ofextracellular matrix, fibronectin and steroidogenic enzymes, and growth ofbovine granulosa cells in anchorage-independent culture. Reprod Fertil Dev8:249–257

40. Lavranos TC, O’Leary PC, Rodgers RJ, Effects of insulin-like growth factorsand binding protein-1 on bovine granulosa cell division in anchorage-inde-pendent culture. J Reprod Fertil 107:221–228

41. Spencer SJ, Cataldo NA, Jaffe RB 1986 Apoptosis in the human reproductivetract. Obstet Gynecol Surv 51:314–323

2612 GRANULOSA CELL DEATH IN BOVINE FOLLICLES Endo • 1999Vol 140 • No 6