Journal of Immunological Methods, 137 (1991) 79-87© 1991 Elsevier Science Publishers B.V. 0022-1759/91/$03.50ADONIS 002217599100096E

JIM 05842

Detection of apoptosis of immature CD4+8+ thymocytesby flow cytometry

79

Wojciech Swat, Leszek Ignatowicz and Pawel KisielowLaboratory of Cellular Immunology, Institute of Immunology and Experimental Therapy, Polish Academy ofSciences, Wroc!aw, Poland

(Received 14 June 1990, revised received 15 October 1990, accepted 5 November 1990)

Apoptosis, i.e., programmed cell death, may be the mechanism by which both autoreactive andunselected immature CD4+ 8+ thymocytes are eliminated in the thymus. In the present paper we describea simple and rapid flow cytometric method which permits one to study the induction and kinetics ofapoptosis of CD4 + 8+ thymocytes using in vivo and in vitro suspension cultures. Analysing the level ofsurface expression of CD4 and CD8 molecules, forward light scatter and side (90 0

) scatter as well asstaining with ethidium bromide, three distinct stages of apoptosis of CD4 + 8+ thymocytes were defined.

By counting cells passing through different stages of apoptosis one can attempt to quantitate thisprocess. This method should be useful for in vitro studies on the mechanisms of negative and positiveselection of CD4+ 8 -I- thymocytes, i.e., induction and inhibition of apoptosis respectively.

Key words: Apoptosis; CD4 + 8+ thymocyte; Flow cytometry

Introduction

Recent evidence suggests that CD4 + 8+

thymocytes represent the critical stage in T celldevelopment at which the specificity of a ran­domly generated a/f3 T cell receptor (TCR) isscreened for self reactivity (Kisielow and VonBoehmer 1990). Experiments using TCR trans­genic mice have suggested that CD4 + 8+ thymo­cytes are deleted when the TCR binds to theMHC molecule plus specific peptide presented bythe stromal, bone marrow derived cells in the

Correspondence to: P. Kisielow, Institute of Immunologyand Experimental Therapy, ul. Czerska 12, 53-114 Wroclaw,Poland.

Abbreviations: TCR, T cell receptor; FSC, forward lightscatter; SSC, side scatter; PE, phycoerythrin; FITC, fluo­rescein isothiocyanate; EB, ethidium bromide; IMDM, Iscovemodified Dulbecco's medium.

thymus (Kisielow et al., 1988a; Sha et al., 1988a;Pircher et aI., 1989), but they are rescued fromdeath and induced to mature when the receptorbinds to the MHC molecule on thymic epitheliumin the absence of specific peptide (Kisielow et al.,1988b; Sha et al., 1988b; Teh et al., 1988; Berg etal., 1989; Kaye et al., 1989; Scott et al., 1989).Experiments using thymus organ cultures treatedwith anti CD3 antibodies (Smith et al., 1989) orwith superantigens (Jenkinson et al., 1989) havesuggested that death of autoreactive CD4 + 8+

thymocytes occurs as a consequence of apoptosis,i.e., programmed cell death, triggered by TCR­mediated transmembrane signalling.

It seems therefore that the elucidation of themechanisms responsible for both the inductionand inhibition of apoptosis in CD4 + 8+ thymo­cytes is crucial for a better understanding of the.selection mechanisms operating in the thymus.The development of techniques to study these

80

mechanisms using CD4+ 8 + thymocytes in sus­pension culture would permit experiments whichcannot be performed in in vivo or in vitro organculture systems. For example the role of thymicstromal cells in the negative and positive selectionof thymocytes could be studied in greater detail.

U sing flow cytometric analysis to detect changesof various cellular parameters of thymocytes cul­tured in suspension we have been able to identifydistinct stages of apoptosis of CD4 + 8+ thymo­cytes. Here we describe this method and show thatit can be used to demonstrate the induction ofapoptosis of CD4 + 8 + thymocytes by stromal cells,in this case thymic macrophages.

Materials and methods

MiceNewborn or young adult (4-6-week-old)

BALB/c mice were obtained from the colony bredand maintained at the Institute of Immunologyand Experimental Therapy, Wroc1aw, Poland.

MediumIscove's modified Dulbecco's medium (IMDM,

Gibeo), supplemented with 5 X 10-4 M 2­mercaptoethanol, penicillin (100 U/ml), strep­tomycin (100 p.g/ml) and 10% fetal calf serum(Flow Laboratories, Irvine, Scotland), was usedfor suspension culture of thymocytes.

ThymocytesSuspensions of thymocytes were prepared in

IMDM by pressing the thymus with a syringeplunger through a fine nylon mesh. After onewashing, cells were cultured as described below.

Cortisone treatmentMice were injected i.p. with hydrocortisone

acetate (Polfa, Poland) using a dose of 4mgy'mouse in 300 J.LI of PBS. Control mice wereinjected with 300 p.l of PBS only.

Thymic macrophagesThe clone D6 was obtained from fetal murine

thymus of BALB/c (H_2d) strain. This clone is

weakly phagocytic, forms rosettes preferentiallywith CD4 + 8+ thymocytes, reacts with anti-Thyl

and anti-Macl antibodies, and expresses a lowlevel of class II MHC antigens, which can beincreased by pretreatment with interferon-v (L.L,W.S. and P.K. manuscript in preparation).

Cell culturesThymocytes (2 X 105/well) were cultured either

alone or together with cells of a D6 clone (2 X

104/well) in flat bottomed tissue culture micro­plates (Costar) for 24 or 48 h at 37 0 C in CO2incubator. To some cultures, ionomycin was ad­ded at a concentration of 5 p.g/ml.

Antibodies and staining of cellsThe rat monoclonal antibodies (fluorescein­

conjugated anti-mouse Lyt-2 (FITC-anti-CD8)and phycoerythrin-conjugated anti-mouse L3T4(PE-anti-CD4) were purchased from Becton Dick­inson (Mountain View, CA) and were used at thepredetermined, optimal concentrations. Stainingand washing of cells was done in cold PBS (4 0 C)containing 5% fetal calf serum (FCS, FlowLaboratories, Irvine, Scotland). For staining, 2 X105 thymocytes were incubated for 30 min at 4 0 Cin the mixture of optimally diluted PE-anti-CD4and FITC-anti-CD8 antibodies in a final volumeof 0.2 ml and then washed twice before analysison a fluorescence activated cell sorter (FACS). Insome experiments, ethidium bromide (EB, 1p.gjml) was added 5 min before analysis in orderto stain dead cells in the sample.

Flow cytometryFlow cytometry and cell sorting were per­

formed using the FACStar apparatus (BectonDickinson) equipped with a single argon ion laserfor FITC (488 nm), PE and EB (480-550 nrn)excitation, interfaced with a HP9000/300 Hew­lett-Packard computer. Due to the non-overlap­ping intensity of signals (see Figs. 3a and 3b) thesame red fluorescence detector was used forsimultaneous analysis of PE and EB staining. Flu­orescence data were collected using logarithmicamplification (with a three-log orders scale) on10,000 cells (both dead and viable), excluding celldebris by a combination of forward light scatter(FSC) and 90 0 scatter (SSC). Cell frequencyhistograms were plotted in 1024 channels on the xaxis (FSC) and cell number shown on y axis.

Two-colour immunofluorescence data were dis­played as dot plots in which log intensities ofgreen (FITC) fluorescence were plotted on the xaxis, and log intensities of red (PE, EB) fluores­cence were plotted on the y axis, in a 64 X 64channel matrix. For acquisition, analysis and elec­tronic gating of data we used the Becton Dickin­son FACStarPLUS programme.

Separation of the CD4highghigh and CD41owgiowthymocytes was performed in the cold using thegates shown on Fig. 1: 1 X 106 cells of each popu-

THYMOCYTES

81

lation were collected into tubes prefilled with icecold FCS and then reanalysed. The purity of thesorted populations was > 95%.

DNA electrophoresisAgarose gel electrophoresis of DNA was per­

formed as described by Smith et al. (1989), withminor modifications. Samples of 4 X 10 5 cells werewashed and pelleted in PBS at 4 0 C. Pellets wereresuspended in 20 III aliquots of DNA buffer (10

a)FRESH AFTER 24 HRS IN CULlURE

l.I'JSTAJNEDCONTROL

1· 50 %2- 17%

1· 7 4"102· 2%

anti ·CDS (Fll C)

L03 GREEN FLUORESCENCE

b)

i . 1 1 .. . . , • • • 1_0 , . 1 ... 1

a:wt:<oen2>m

FORWARDSCATIERFig. 1. Phenotype of in vitro cultured thymocytes. Thymocyte suspensions were prepared and analysed for expression of CD4 andCD8 molecules (a) and for side vs. forward light scatter (b) either before (left panels) or after (right panels) 24 h culture. Populationsexpressing high and low levels of CD4/CD8 molecules (a) as well as having bigger size/lower dens ity and smaller size/ higher

density (b ) are indicated by arbitrary gates.

volume and increased density (Fig. Ib). A popula­tion of CD4+ 8+ thymocytes expressing reducedlevels of CD4 and CD8 molecules would be ex­pected to exhibit these characteristics.

82

mM EDTA, 50 mM Tris HCI (pH 8.0) containing0.5% (wjv) sodium lauryl sarcosinate (BRL) and0.5 mgy'ml proteinase K (Sigma» and incubatedat 50 0 C for 30 min. After incubation 10 p.l of 0.5mgy'ml RNase A (Sigma) in DNA buffer wereadded to each sample, and incubation continuedfor a further 30 min. Samples were then heated to70 0 C and 5 I.d of gel-loading buffer (10 mMEDTA pH 8.0 containing 0.25% bromophenol blueand 40% (wIv) sucrose) were mixed with eachsample before loading into the dry wells of 1.8%(wjv) agarose gel containing 0.5 fLgjml ethidiumbromide. Electrophoresis was carried out in 100mM Tris-acetate buffer pH 7.8 until the markerdye had migrated for a distance of 6-8 cm.

Results

Detection of two subpopulations of CD4 + 8 + cellsamong cultured thymocytes

Fig. 1a shows the results of direct double stain­ing of fresh and in vitro cultured thymocyte sus­pensions with anti-CD4 and anti-CD8 antibodies(the recovery of cells after 24 h in culture wasabout 50-80% of the input). While staining offresh thymocytes revealed a typical picture of threemajor subpopulations (CD4, CD8 double positive,CD4 single positive, and CD8 single positive),staining of thymocytes cultured in suspension for24 h clearly identified an additional subpopulationexpressing reduced levels of CD4 and CD8 mole­cules. Moreover in the plots of FSC (indicatingcell volume) versus SSC (indicating cell density),an additional population accumulating among thecultured thymocytes comprised cells of decreased

Fig. 2. Agarose gel analysis of DNA from cultured thymocytesshowing degradation into oligonucleosomal fragments of DNAfrom CD4 + g+ thymocytes expressing low level of CD4 andCD8 molecules. CD4highghigh Cupper') and CD41ow81ow('lower') populations of cultured CD4 +8+ thymocytes, shownin Fig. I, were sorted out on FACStar. Lane 1: DNA isolatedfrom unseparated thymocytes; lane 2: DNA from sorted' lower'population of thymocytes; lane 3: DNA from sorted 'upper'

population of CD4+ g + thymocytes.

1 2 3

a)

DISPLAY OFGATES :

b)

83

1- 62% W2- 27%

~3- 0·.4 CD

ffi w+

~we:.~

Cl Clw

~a:

§ c:C1l

1- 60 %2- 16 %3- 17%

d)c) UNGATED (b)

anti-CDS (FITC)

L03 GREEN FLUORESCENCE

GATE 1(b)

a:w

Soen&en

e) GATE 2(b)FORWARDSCADER

f) GATE 3(b)

a:w

Soen&en

FORWARD SCADER

Fig. 3. Multiparameter flow cytometric analysis of cultured thymocytes identifying distinct stages of ap optosis of CD4 +8 + cells.Ethidium bromide was added to a samp le of 24 h cultured thymocytes stained by anti-C D4 and anti-CDS antibodies ( a ) andreanalysed (b) . Panels c- f demonstrate the side vs. forward light scatter plots of populations which were electronically gated asshown in panel b. The CD41owSIow population, unstained by EB (gate 2, panel b) contains cells passing through first two stages ofapoptosis. Stage 1 involves reduction of expression of CD4/CDS molecules without changing cell volume and density (population 1,panel e), stage 2 involves in addition a decrease in volume and an increase in density (population 2, panel e) . The CD41ow81owpopulation, stained by EB (gate 3, panel b) represents the third stage of apoptos is and involves a further decrease in cell volume and

an increase in cell density (population 3, panel f) .

84

CD4 + 8 + thymocytes expressing low level of CD4and CD8 molecules undergo apoptosis

Loss of volume and increase of density arecharacteristic features accompanying apoptosis,distinguishing this pathway of death from necrosis(Wyllie, 1988). To confirm that it was the sub­population of CD4+8+ thymocytes with reducedexpression of CD4/CD8 molecules which was un­dergoing apoptosis we sorted out 'upper' (gate 1,Fig. 1) and 'lower' (gate 2, Fig. 1) populations ofdouble positive thymocytes and performed agarosegel electrophoresis of DNA isolated from them.As shown in Fig. 2, DNA from the 'lower' popu­lation was degraded whereas DNA from the' up-

per' population was not, indicating that CD4 +8+

thymocytes expressing reduced level of CD4 andCD8 molecules were dying by apoptosis.

Identification of distinct stages of apoptosisThe addition of ethidium bromide (EB), which

marks dead cells, does not stain all cells from the'lower' population of cultured CD4 +8+ thymo­cytes (Fig. 3b) suggesting that this population isnot homogeneous and contains cells at differentstages of apoptosis. Note that the addition of EBto fresh, uncultured suspension of thymocytes doesnot stain significant proportion of cells (see Fig.5). As shown in Figs. 3c-3f, several distinct stages

w

~cr@ex:§

Wa.

a)

".: .. " \.':".... :

b)

anti-COB (FITC )

LOO GREENFLUORESCENCE

c)

ex:w1=«oC/)

g,C'l

FORWARD SCAlTER

• • "' ' 0

Fig. 4. Induction of apoptosis of CD4 + 8 + thymocytes in suspension culture by a thymic stromal cell line. Unseparated suspensionsof thymocytes were cultured for 24 h either alone (a) or in the presence of thymic macrophages (D6 cell line, thymocyte/D6 cellratio = 10/1) (b) or ionomycin (5 JLg/ml) (c) and analysed for expression of CD4 and CD8 by two colour immunofluorescence (top)

and for changes in FSC/SSC values (bottom).

85

of apoptosis can be defined by different positionson the FSC versus SSC plots. The viable cellsfrom the 'upper' CD4 + 8+ population are thelargest and least dense (Fig. 3d ). The majority of' lower' CD4 + 8+ thymocytes not stained by EBare smaller and more dense but some of them donot yet differ in size and density from ' upper'CD4+ 8+ thymocytes (Fig. 3e). Cells which do notstain with EB are probably at the initial stage ofapoptosis during which internucleosomal DNA

cleavage already occurs but the cytoplasmic mem­brane is still in tact and impermeable to EB.CD4 '"8+ thymocytes stained by EB, probablyrepresenting cells at the latest stage of apoptosis,are the smallest and their density is the highest(Fig. 3f). Thus, analysing the above parameters,one can distinguish at least three stages of apopto­sis which can be quantitated by counting cells indifferent compartments at different time points(see Fig. 5).

TIMEAFTERHYOROCORT1SO. E INJECTION :

o 12 24 4e (HAS)

\.2­3 ·

t - 1%2· 3%

eJ·····- t o '.... .\ . \0%2. 7%3· 16%

j , 10%2· \2%

\. 29 %2- 24%

tf).:.r-,\.

•• . » :

" , ~:. ~

j , 71%2· 5%

ill!!:..

C5o.~

~'"

() \- 70%

~... -. 2· 5%

3- 2%

~ ccW 1 ,

§ +ill!!:....,0~.~

'"anti·COe (FITC)

LOOGREEN~ESCENCE

Fig. 5. Flow cytometric detection of in vivo induction of apoptosis of CD4 + 8+ thyrnocytes by hydr ocortisone. Thym ocytesuspensions were prepared at several timepoints after injec tion of hydrocor tisone, stained with PE-Iabelled anti-CD4 and FITC­labelled anti-CD8 antibody in the presence (bottom) or absence (top) of ethidium bromide and analysed for two colour fluorescence.The results of agarose gel electrophoresis of DNA isolated from the unseparated populat ion of thymocytes obtained at different

timepoints after hydrocor tisone injection are also shown.

86

Detection of apoptosis of CD4 +8 + thymocytes in­duced by a thymic stromal cell line

Having established a method for detecting'spontaneous' apoptosis of CD4+S+ thymocytesin suspension culture we were next interested tofind out whether one could apply this method tostudy the induction of apoptosis by thymic stromalcells. Fig. 4 shows that after 24 h of co-culturewith thymic macrophages (D6 clone) there was asignificant and specific depletion of cells in the'upper' population and a proportional increase inthe number of cells in the 'lower' population ofCD4 +8+ thymocytes. The populations of doublenegative (CD4-8-) and single positive (CD4+S-,CD4 - 8+) thymocytes remained unaffected. Asimilar effect was seen in the presence of murinebut not guinea pig peritoneal macrophages (datanot shown) and in the presence of ionomycin (Fig.4), a known inducer of apoptosis in CD4 +8+thymocytes (Smith et al., 1989).The DNA isolatedfrom CD4low8low thymocytes of ionomycin treatedand D6 clone co-cultured samples was degradedinto oligonucleosomal fragments (data not shown).Co-culture with other murine cells tested such asT and B cell lymphomas and fibroblast lines waswithout any visible effect on the thymocytes (datanot shown).

Detection of apoptosis of CD4 +8 + thymocytes in­duced by hydrocortisone in vivo

Fig. 5 shows results which demonstrate that thepresent method may also be applied to monitorthe kinetics of apoptosis induced by the in vivoadministration of glucocorticoids (Wyllie andMorris, 1982). After i.p. injection of hydrocorti­sone the gradual disappearance of viable CD4 +S+thymocytes, which pass through the above de­scribed stages of apoptosis, was observed.

Discussion

In the present study we have shown that murineCD4 +8+ thymocytes expressing lower than nor­mal levels of CD4 and CD8 molecules represent asubpopulation of immature thymocytes undergo­ing programmed cell death, i.e., apoptosis. Thecrucial evidence for this conclusion comes fromthe demonstration of almost complete degradation

of DNA isolated from this population. Analysingvarious cellular parameters by flow cytometry, wewere able to distinguish three distinct stages ofapoptosis of CD4 +8+ thymocytes (Fig. 3). Ourdata suggest the following interpretation. In thefirst stage only reduced expression of CD4 andCD8 molecules is observed and in the next stagethis is followed by an increase of cell density witha concomitant decrease in cell volume. In thissecond stage cells are still refractory to stainingwith DNA stains such as EB or propidium iodide.Staining with EB is characteristic of the thirdstage of apoptosis which is accompanied by afurther decrease in cell volume and an increase ofcell density. Enumeration of cells found at differ­ent stages of apoptosis at different time pointsshould permit quantitation of the process.

We believe that the ability to study the processof apoptosis of CD4 +8+ thymocytes in suspen­sion culture will lead to a useful in vitro systemsuitable for detailed studies of the conditions re­sponsible for the positive or negative selection ofimmature thymocytes. In a preliminary attempt toestablish such a system, we have used a clone of athymic macrophage cell line (D6) to demonstratethe ability of stromal cells to accelerate the pro­cess of apoptosis occuring 'spontaneously' in sus­pension culture. The results indicated that D6cells are able to induce apoptosis of CD4 +8+thymocytes in suspension culture (Fig. 4). Themechanism and specificity of this effect is cur­rently under investigation. The process of apopto­sis, induced by D6 cells, appears to be TCR-inde­pendent and therefore other activating accessorymolecules may be involved.

The formation of CD4 +8+ thymocytes ex­pressing reduced levels of CD4 and CDS mole­cules after in vitro culture has been observedrecently by various investigators (Inaba et al.,1988; Kyewsky et a1., 1989), but none of thesestudies identified these cells as apoptotic. On thecontrary, they were assumed to represent a viablepopulation of cells and the degree of cellular DNAdegradation was not tested (Inaba et al., 1988;Nakashima et al., 1990). In the studies of Inabaand colleagues (Inaba et a1., 1988) it was claimedthat the CD41owCD81ow and not the CD4 high

CDShigh thymocytes are recognized, killed andcleared by macrophages. Our results presented in

this paper are in conflict with their work. Ourresults suggest that apoptosis-undergoing CD4 10w

CDS low cells are derived from the CD4highCDShighpopulation during co-culture with macrophages.

Acknowledgements

We would like to thank Ewa Ziolo and JaninaCzech for expert technical assistance , AleksandraMoniewska for help with FACS operations andDr. Harald von Boehmer for discussion and criti­cal reading of the manuscript.

This work was supported by Grant 10.5 fromthe Polish Academy of Sciences.

References

Berg, L.J., Pullen, A.M., Fazekas de St. Groth, B., Mathis, D.,Benoist, e. and Davis, M.M. (1989) Antigen /MHC-specificT cells are preferentially exported from the thymus in thepresence of their MHC ligand. Cell 58, 1035.

Inaba, K., Inaba, M., Kinashi , T., Tashiro, K., Witmer-Pack,M., Crowley, M., Kaplan, G., Valinsky, J., Romani, N.,Ikehara, S., Muramatsu, S., Honjo, T. and Steinman, R.M.(1988) Macrophages phagocytose thymic lymphocytes withproductively rearranged T cell receptor alpha and betagenes. J. Exp. Med. 168, 2279.

Jenkinson, EJ., Kingston, R., Smith, e.A., Williams, G.T. andOwen, JJ.T. (1989) Antigen-induced apoptos is in develop­ing T cells: a mechanism for negative selection of the T cellrecep tor repertoire. Eur. J. Immuno!. 19, 2175.

Kaye, J., Hsu, M.L., Sauron , M.E., Jameson , J.C., Gascoigne,R.J. and Hedrick, S.M. (1989) Selective development ofCD4 T cell in transgenic mice expressing a class II MHC­restricted antigen receptor . Nature 341, 746.

Kisielow, P. and Von Boehmer, H. (1990) Negative and posi­tive selection of immature thymocytes: timing and the roleof the ligand for alpha/beta T cell receptor. Semin. Irn­muno!. 2, 35-44.

87

Kisielow, P., BlUthmann, H., Staerz, U., Steinmetz, M. andVon Boehmer, H. (1988a) Tolerance in T cell receptortransgenic mice involves deletion of nonmature CD4+ 8+thymocytes. Nature 333, 742.

Kisielow, P., Teh, H.S., Bluthmann, H. and Von Boehmer, H.(1988b) Positive selection of antigen specific T cells inthymus by restricting MHC molecules. Nature 335, 730.

Kyewsky, B.A., Schirrmacher, V. and Allison, J.P. (1989) Anti­bodies against the T cell receptor/CD3 complex interferewith distinct intra-thymic cell-cell interactions in vivo: cor­relation with arest of T cell differentiation. Eur. J. Im­muno!' 19, 857.

Nakashima, M., Mori, K., Maeda, K., Kishi, H., Hirata, K.,Kawabuchi, M. and Watanabe, T. (1990) Selective elimina­tion of double-positive immature thymocytes by a thymicepithelial cell line. Eur. J. Immunol. 20, 47.

Pircher, H., Burki, K., Lang, R., Hengartner , H. andZinkernagel, R.M. (1989) Tolerance induction in doublespecific T-cell receptor transgenic mice varies with antigen.Nature 342, 559.

Scott, B., B1uthmann, H., Teh, H.S. and Von Boehmer, H.(1989)The generation of mature T cells requires interactionof the alphajbeta T cell receptor with major histocompati­bility antigens. Nature 338, 591.

Sha, W.e., Nelson, e.A., Newberry, R.D., Kranz, D.M., Rus­sel, J.H. and Loh, D.Y. (I988a) Selectiveexpression of anantigen receptor on CD8-bearing T lymphocytes in trans­genic mice. Nature 335, 271.

Sha, W.e., Nelson, e.A., Newberry, R.D., Kranz, D.M., Rus­sel, J.H . and Loh, D.Y. (1988b) Positive and negativeselection of an antigen receptor on T cells in transgenicmice. Nature 336, 73.

Smith, c.x., Williams, G.T., Kingston, R., Jenkinson, EJ. andOwen, J.J.T. (1989) Antibodies to CD3/T-cell receptorcomplex induce death by apoptosis in immature T cells inthymic cultures. Nature 337, 181.

Wyllie, A.H. (1988) Apoptosis. lSI Atlas of Science: Immunol­ogy I, 192.

Wyllie, A.H., Morris, R.G. (1982) Hormone-induced celldeath.Purification and properties of thymocytes undergoingapoptosis after glucocorticoid treatment. Am. J. Pathol.109, 78.