Dexamethasone Induces Irreversible GIArrest and Death of a Human Lymphoid Cell Line JEFFREY M. HARMON,' MICHAEL R. NORMAN,' BETTY JO FOWLKES AND E. BRAD THOMPSON ' ' Laboratory of Biochemistry and Laboratory of Pathology, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20014; Department of Chemical Pathology, Kings College Hospital, London, England
ABSTRACT
Growth of a human leukemic T-cell line (CEM C7) in M dexamethasone results in inhibition of growth and rapid loss of cell viability after a delay of approximately 18 to 24 hours. Analysis of dexamethasonetreated cells by flow-microfluorometry showed that they were arrested in the GI phase of t h e cell cycle. Loss of cell viability began at the same time as G, accumulation was first detectable, and 20% of all cells were found to be blocked in G, at this time suggesting that loss of viability and GI arrest were coincident events. Half-maximal and maximal effects on both viability and GI arrest after 48 hours in steroid were nearly identical with respect to steroid concentration and corresponded to half-maximal and full occupancy of glucocorticoid specific receptor by hormone, consistent with a glucocorticoid receptor mediated mechanism for both phenomena. Most non-viable cells were arrested in G1,and accumulation of cells in G1 was irreversible; removal of steroid in the presence of colcemid did not result in a decreased fraction of G, cells. Furthermore, dexamethasone treatment did not protect cells against the effects of 33258 Hoechstamplified killing of bromodeoxyuridine substituted cells exposed to light. These results show t h a t dexamethasone arrests these leukemic cells in GI and strongly suggest t h a t dexamethasone-treated cells are killed upon entry into GI.
'
The multiple effects of glucocorticoid hormones on lymphoid tissue have been widely studied and documented. Inhibitory effects on a number of cellular processes have been described including those on transport of glucose and some amino acids (Rosen et al., '72; Makman et al., '681, on incorporation of precursors into RNA and DNA (Makman et al., '70; Hofert and White, '68),and ultimately on cell growth. Glucocorticoids cause lysis of rodent thymus cells and thymus-derived cell lines both in vivo (Dougherty and White, '45; Claman, '72), and in vitro (Harris, '70; Bourgeois and Newby, '77). Lysis of untransformed thymus cells has been difficult to study in vitro due to the relatively short period of time during which t h e tissue or dissociated cells can be maintained in a viable state. Consequently, several rodent lymphoid tissue cult u r e lines have been developed to facilitate such in vitro studies (Harris, '70; Horibata J. CELL. PHYSIOL. (1979) 98: 267-278.
and Harris, '70). We have recently characterized a clonal human lymphoblastoid cell line (CEM-C7) derived from the CCRF-CEM line, which was originally established from a patient with acute lymphoblastic leukemia. The CEM-C7 line is sensitive to glucocorticoid hormones and is the first human line in which glucocorticoid receptor occupancy has been correlated with the growth inhibitory effects of the steroid (Norman and Thompson, '77). In this sense the CEM cell line appears analogous to t h e S49 mouse lymphoma line, from which several classes of steroid resistant mutants have already been isolated (Sibley and Tomkins, '74; Yamamoto et al., '74). The mechanism of the inhibitory effects of Received Apr. I, '78.Accepted Aug. 30, '78. ' Permanent address: Department of Chemical Pathology, Kings College Hospital, Landon, England. Present address: Laboratory of Microbial Immunity, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, Maryland 20014.
267
260
HARMON, NORMAN, FOWLKES AND THOMPSON
glucocorticoids on cell growth is still unknown. However, the possibility that glucocorticoids produce their growth inhibitory effects on lymphoid cells by blocking them in the G, phase of the cell cycle was suggested by studies in which patients with acute lymphoblastic leukemia (ALL) were treated with glucocorticoids. Lymphoblasts from these patients were found to have a decreased 3H-thymidine labeling index several hours before a decrease in mitotic index was observed (Ernst and Killman, '70; Lampkin et al., '71). We felt that it was important to determine whether steroids indeed block leukemic cells a t a specific point in the cell cycle. If so, it would have important implications for mechanisms of cell killing. The reversibility, or irreversibility of the block would have important pharmacologic implications, inasmuch as two reversible cycle-active drugs used in combination during chemotherapy might interfere with each other. If the effect were reversible, it would also provide unique advantages for isolation of conditional steroid resistant variants. We therefore have used the CEM-C7 cell line to study the lethal effects of glucocorticoids on human T cells. In particular we have found that steroid treatment of these cells results in an irreversible accumulation of cells in G,.
loid fibroblasts (CRL. 1221, American Type Culture Collection) was seeded onto the desired number of 60-mm plates. Just prior to the plating of cells a 2.5-ml separator layer of 10% FCS containing 0.25% RMPI 1640 agarose (Type 11, Sigma Chemical Co.) maintained a t 42°C was pipetted onto each plate and allowed to gel. Immediately thereafter, 1 volume of a cell suspension of appropriate density was mixed with 4.5 volumes of RPMI 1640 a t twice isotonic concentration, containing 20% FCS a t 37°C. To this suspension 4.5 volumes of 0.5% agarose a t 42°C were ,quickly added and 2.5 ml of this mixture were rapidly pipetted over each separator layer and allowed to gel a t room temperature. Plates were incubated for two to three weeks, stained with neutral red and the number of colonies determined. All colonies containing more than 100 cells were counted. In separate experiments efficiency of colony formation varied between 40 and 60%.Within a single experiment, however, cloning efficiency of multiple control cultures was constant.
+
Flow-microfluorometry
Cell cycle analyses were obtained using the fluorochrome, mithramycin (Pfizer), the cellular binding of which has been shown to be stoichiometrically related to DNA content MATERIALS AND METHODS (Crissman and Tobey, '74). Cultured cells were washed two times with phosphate buffered Cell growth and culture saline (PBS: KC1, 0.2 glliter; KH2P04,0.2 g/ The clonal cell line CEM-C7 was cloned from liter; NaC1, 8.0 glliter; and NaZHPO4,2.16 g/ the cell line CCRF-CEM as previously de- liter) and fixed in 70%ethanol. The fixed cells scribed (Norman and Thompson, '77). Cells were resuspended in 3-5 ml of staining soluwere grown in RPMI 1640 (NIH Media Unit) tion (0.85%NaCl solution containing 15 mM containing 10% heat inactivated fetal calf MgClz and 100 pg/ml mithramycin) for a t serum (North American Biologicals).Cultures least 20 minutes a t room temperature. Cells were maintained a t densities of 5 x lo4- 2 x were filtered through a 62-p nylon mesh to re106per ml a t 37°C in a humidified atmosphere move any clumps. The concentration of cells of 95%air, 5%COz,as stationary suspensions. in staining solution was adjusted to approxUnder these conditions repeated determina- imately 4-6 X 105/mlto minimize coincidence tions showed a population doubling time of 18 errors during analysis. to 20 hours. Cell number was determined usThe DNA content of individual cells was aning a Model B Coulter counter (Coulter Elec- alyzed by measuring the mithramycin stained tronics). The lower threshold was set so that cells using the Los Alamos Scientific Laborawhen cells were treated with dexamethasone, tory (LASL) Multiparameter Analyzer and pyknotic cells would not be excluded. Cell Sorter. The design and operation of this instrument has been described by Steinkamp Determination of cell viability et al. ('73). A 457-nm argon ion laser line was Cell viability was assayed by determining used for excitation, and total fluorescent the colony-forming efficiency of single cells emission above 495 nm was measured. Histoplated in semisolid medium. Several days grams of the DNA spectra were produced by a before each experiment a layer of human dip- DEC PDP 11/40 Computer (Maynard, Massa-
269
G, ARREST AND KILLING OF CELLS BY DEXAMETHASONE
chusetts) with an RT 11 Operating System. The data analysis, storage, and graphics used were as described by Miller e t al. (’78). The determination of DNA content on individual cells yields a distinctive distribution spectrum (cf. fig. 2). The spectrum shows two peaks; the first, that of cells with a G, content of DNA, the second t h a t of cells with double t h e G, content; that is cells in G Por M phase. The interval between the two peaks corresponds to cells in S phase. This distribution was then analyzed according to the method of Miller et al. (‘78) to derive the percentage of cells in GI, S, and Gz M. Compartment boundaries were selected for integration using control samples and were maintained for all subsequent analyses. One hundred thousand cells were analyzed t o generate each histogram. Calibration of instrument for linearity and coefficient of variation using standard particles (Particle Technology, Inc., Los Alamos, New Mexico) was always done prior to analyzing samples. The coefficient of variation of these particles was maintained between 4.04.5%throughout these studies. 33258 Hoechst-amplified BrdUrd-light killing Cells were photosensitized essentially by t h e method of Stetten et al. (‘76). Cultures were grown in the presence of BrdUrd (5-bromodeoxyuridine, Sigma) for 24 hours. Freshly prepared 33258 Hoechst (Riedel-DE Haen Ag Seelze, Hannover) was added to the culture to give a final concentration of 1.5pg/ml and the cells returned to the incubator for a n additional three hours. Cells to be irradiated were transferred to fresh medium and 2.5 ml of cell suspension placed into 60-mm covered tissue culture dishes. Dishes were exposed from above to two Sylvania Cool White fluorescent lamps (F15 T 8 - CW, Sylvania) at a distance of 6 cm for up to five minutes. Irradiated cells were plated at 250, lo3, and lo4cells/plate in quadruplicate and plates with 20 to 200 colonies per plate were counted as described above. Fractional survival was determined as the percent colony formation of the treated cells divided by t h e percent colony formation of a n untreated control.
I
+
,
t
RESULTS
Effects of dexamethasone on CEM-C7 cells Addition of dexamethasone to logarith-
1
2
3
4
I 1
I 2
I 3
4
0.
0.0.
I
DAYS Fig. 1 Inhibition of cell growth and viability by dexaM (0- O ) , M methasone. Cells were grown in (0-U), M (A-A) or no (0-0)dexamethasone and periodically assayed for cell number (upper panel) or cell viability (lower panel) as described in MATERIALS AND METHODS.
mically growing cultures of CEM-C7 results in inhibition of cell proliferation after a delay of about 24 hours (fig. 1). This inhibition is accompanied by a progressive loss of cell viability as measured by the loss of plating efficiency in agarose gels (fig. 1). The assumption that the ability to form clones measures
270
HARMON, NORMAN, FOWLKES AND THOMPSON
S
14.5r
- 17.0
G,+M
a
10.9
12.8
I
N
X
I
CHANNEL NUMBER Fig. 2 Influence of dexamethasone on cell cycle distribution. Cells were grown in the presence (b) or absence (a) of dexamethasone for 48 hours and then analyzed by flow-microfluorometryas described in MATERIALS AND METHODS. Abscissa: Channel No. is linear function of DNA/cell. Ordinate: A total of lo5 cells were counted for each histogram and scale difference in the two panels reflects the fact that display of data is normalized to give a uniform maximum peak height.
viability derives from the fact that glucocorticoids cause lysis of CEM-C7 cells in mass culture (Norman and Thompson, '77) and that direct observation of the cells cloned in agarose after steroid treatment showed no intact cells or abortive colonies but only cell debris vs surviving colonies. To ascertain whether these effects were accompanied by cell cycle specific inhibition, the cell cycle distributions of dexamethasone-treated and untreated cultures were studied by flow-microfluorometry (FMF). The results (fig. 2) showed that after 48 hours of exposure to M dexamethasone a far greater proportion of cells had a DNA content characteristic of G, cells than did those of an untreated culture. No clumping of cells was observed in either the steroid treated or untreated cultures. Therefore differential filtration through the 62-pnylon mesh during preparation of cells for FMF cannot explain this result.
Relationship of killing to cell cycle perturbation To determine if the shift in the cell cycle distribution and loss of cell viability were related, the time course of both loss of viability and alteration in cell cycle distribution was observed. Cells were treated with M dexamethasone and a t various times thereafter removed from the steroid, plated for viability, and analyzed by FMF. The results (fig. 3) show that cells begin to lose viability 24 hours after exposure to the steroid. Following this initial delay, loss of viability proceeds rapidly, with a rate of loss of about 50%per six hours. A shift in cell cycle distribution is detectable 18-24 hours after exposure to steroid and cells continue to accumulate in G , thereafter, suggesting a possible temporal relationship between loss of viability and cell cycle perturbation. It was not possible to study the cell cycle distribution beyond 60 hours of steroid treatment
271
G I ARREST AND KILLING OF CELLS BY DEXAMETHASONE
due to cell lysis and accumulation of cellular debris which produced FMF patterns impossible to interpret. One of the characteristics expected for a steroid specific response is that the magnitude of the response be related to the fraction of occupied steroid receptor molecules and therefore be saturable. To determine if loss of cell viability and altered cell cycle distribution met this criterion the dose responsiveness of both loss of viability and cell cycle perturbation was determined. Cultures of CEM-C7 were incubated with various concentrations of dexamethasone for 48 hours and the cells assayed for both viability and cell cycle distribution. The results (fig. 4) show that both 50% loss of viability and 50%of the maximal effect on the cell cycle are obtained in the presence of 2-5 X lo-’ M dexamethasone. This corresponds closely to the previously measured & for the steroid receptor in these cells of 1.4 x lo-’ M (Norman and Thompson, ’77).The maximal effects on both viability and Gl shift ocM dexamethacur a t approximately 5 X sone which is in good agreement with the concentration a t which receptors are saturated (Norman and Thompson, ’77).
Nature o f the cell cycle effect The shift in the cell cycle distribution of dexamethasone-treated cells to one apparently enriched in GI cells could have been the result of an actual accumulation of cells in the G1 phase of the cell cycle or of differential lysis of cells in the other phases of the cell cycle thereby establishing a new distribution. If the latter possibility were the case then cells which are in G, should be able to enter S; whereas if cells are actually accumulating in G1 they should be unable to do so. To distinguish between these possibilities, a population M dexamethasone of cells treated with for 24 hours was exposed to the mitotic blocker colcemid (0.5 pg/ml) for an additional 12 hours in the continued presence of dexamethasone. When the cells were then analyzed by FMF the results showed that while some decrease in the fraction of cells in G1 occurred during the colcemid treatment, approximately 60% of the cells which were in GI when the colcemid was added were still there 12 hours later (table 1comparing lines 3 and 4). Phase contrast microscopy of the culture a t this time revealed no significant cell lysis. Since colcemid treatment of control cells re-
sulted in an almost complete depletion of cells from GI (table 1,line 2), we conclude that the effect of dexamethasone on the cell cycle is to cause an accumulation of cells in G,. In addiM tion, the increase in the fraction of G, cells during the combined treatment indicates that those cells not arrested in G, continue to cycle (table 1,lines 3 and 4). The accumulation of cells in G, could result from two mechanisms. Either cells which have already lost colony forming ability manage to “coast” through the cell cycle until they reach GI or cells lose this ability when they actually enter G,. The former possibility would predict
+
loo
r
I
CELLS IN GI
12
0.01
24
3 6 4 8 6 0
k 12
24
36
48
60
TIME (hrs) Fig. 3 Time course of loss of viability and G Iaccumulation. Cells were grown in M dexamethasone for various lengths of time and then assayed for cell cycle distribution (upper panel) and viability flower panel) as de-
scribed in MATERIALS AND METHODS.
272
HARMON, NORMAN, FOWLKES AND THOMPSON
100'
80
2
>a l-
a
240 UI n
20
0
10-8
I
I 5~10-~
2x10-8
I
1o
I 2x10-7
-~
I 5x10-7
10-6'-
DEXAMETHASONE (M) Fig. 4 Effect of dexamethasone concentration on viability and cell cycle distribution. Cells were grown for 48 hours in various concentrations of dexamethasone and then analyzed for cell cycle distribution ( A ) and
viability (0).
TABLE 1
Early effect of dezamethasone on G Iaccumulation Distribution of cells (%)
Treatment Dexamethasone
Colcemid
-
-
-
12hours
24 hours 36 hours
12hours
-
G,
S
G,+M
46 6 49 31
26 21 13 18
28 14
38 51
M dexamethaaone. A culture of logarithmically growing CEM-C7 cells was divided in half. One-half was incubated with Twenty-four hours later a sample of each was analyzed by FMF (lines 1and 3). The control culture was again divided and colcemid was added to one-half. Colcemid was also added to the culture containing dexamethasone. Twelve hours later, each culture was again analyzed by FMF (lines 2 and 4). The control culture gave t h e same distribution as a t 24 hours ' Determined a s described In MATERIALS AND METHODS. One-half microgramfmilliliter for terminal 1 2 hours.
that loss of viability should precede cell cycle arrest, while the latter would predict that loss of cell viability should either follow or be coincident with cell cycle arrest. Comparison of the curves in the upper and lower panels of figure 3 shows that the onset of G1accumulation and initial loss of viability correspond closely in time. However, the quantity of cells in GI shown in figure 3 cannot be used for comparison since an altered cell cycle distribution becomes apparent only after cells distal to a cell cycle specific block have passed out of the
phase of the cycle in which the block occurs. Thus, early or mid-G, arrest would not be reflected in an altered FMF distribution for several hours after the first cells were arrested. A more direct assessment of the fraction of cells arrested in G, at a specific time may be obtained by adding colcemid to cells at the same time as steroid is removed from the medium. Subsequent growth of the cells will allow unaffected cells to exit G1,leaving only arrested cells to be detected by FMF analysis. Such an experiment was performed on cells grown in
273
G I ARREST AND KILLING OF CELLS BY DEXAMETHASONE
20
I
a
15
16.7
cv
c'!
z
z
X
X
2
5
10
J
-1
0
u
11.1
w
w
5
8.9
8
6.7
6
cv
c'!
z
2
X
y
I
X
4.4
Y 4 -I
J
w
W 0
0
2.2
2
4 . I
200
I
I
400
I
I
200 CHANNEL NUMBER
I
I
400
I
600
Fig. 5 G , arrested cells after 24 hours in dexamethasone. Cells were grown in the absence (a, b) or presence (c, d) of dexamethasone for 24 hours. Dexamethasone was then removed and 0.5 pg/ml colcemid was added to one untreated (b) and one steroid treated (d) culture. Growth was continued for 12 hours a t which time cells were analyzed by FMF as described in MATERIALS AND METHODS.
M dexamethasone for 24 hours. Twelve hours of colcemid treatment obliterates the characteristic G , peak of control cells (fig. 5b). However, colcemid treatment of cells exposed M dexamethasone for 24 hours leaves a to
significant fraction of cells in G , (fig. 5d). This fraction, about 20% of the entire population, therefore seems to be arrested in G I . This result is in good agreement with that of the experiment of figure 3, which showed a compara-
274
HARMON, NORMAN, FOWLKES AND THOMPSON
ble loss of viability after 24-hour treatment with dexamethasone. These results clearly show that loss of cell viability does not precede cell cycle arrest. The possibility that the GI accumulation is the result of cells which have been killed elsewhere but have managed to reach G , is thus excluded.
o CONTROL 0
HOECHST
A BrdUrd
A BrdUrd
0.001
I
I 1
1
I
2
3
I 4
+ HOECHST
I 5
MINUTES IRRADIATION
Fig. 6 Photosensitization of BrdUrd substituted cells by 33258 Hoechst. Parallel cultures (4) were grown in the presence or absence of 2 X lo-' M BrdUrd for 24 hours. 33258 Hoechst (1.5 pg/ml) was added to one of the BrdUrd substituted and one of the unsubstituted cultures and all four cultures were incubated for an additional three hours. Cells were resuspended in fresh medium and irradiated and assayed for viability as described in MATERIALS O ) , 3 hours AND METHODS. (0-01,no addition; (0in 33258 Hoechst; (A-A), 24 hours in BrdUrd; (A-A), 24 hours BrdUrd, 3 hours 33258 Hoechst.
Irreversibility of cell cycle effect To determine if cell cycle arrest by dexamethasone was a reversible effect two parallel M dexamethacultures were incubated in sone for 48 hours, after which the cells were resuspended in fresh medium. One of the culM dexatures was resupplemented with methasone and both cultures were incubated for an additional 12 hours. Cells from both cultures and a control, untreated culture were then analyzed by FMF. The results of this experiment are presented in table 2. The absence of dexamethasone during the final 12 hours of incubation resulted in no significant reversal of the G, accumulation. Although such a reversal may occur in the small fraction of surviving cells (figs. 1, 3), this would be difficult to detect since this fraction (-7%) represents only a small proportion of the total number of cells analyzed. However, 95%,not 75% of the cells are incapable of forming colonies after 60 hours of dexamethasone (fig. 3, table 2). Therefore there are many cells outside of GI which are not capable of forming colonies. The inability of these cycling cells to form colonies when steroid is removed most likely indicates that these cells have sustained insufficient damage in G I to prevent a limited amount of cycling but are sufficiently damaged to prevent the degree of proliferation necessary to clone in semisolid medium. This view is supported by microscopic examination of dexamethasone-treated cells at this time which shows virtually all of them to be severely deformed (Norman and Thompson, '77). To determine whether the small steroid surviving fraction is composed of cells reversibly arrested in GI we employed a second approach. Cells blocked in GI should incorporate sharply reduced amounts of BrdUrd into their DNA. Such cells should therefore be less sensitive to the lethal effects of incubation in 33258 Hoechst dye followed by exposure to light
TABLE 2
Reversibility of G Iaccumulation Distribution of cells (%)
n m e of treatment
I
'
0-48
48-60
G,
S
G,+M
-
-
Dexamethasone Dexamethasone
Dexamethasone
55 76 15
22 12 14
23 12 11
-
All determinations were made at 60 hours.
G, ARREST AND KILLING OF CELLS BY DEXAMETHASONE
275
loss of cell viability. Five minutes of irradiation of such BrdUrd-treated cells results in a 70%loss of viability. However, with cells ideno CONTROL tically incubated with BrdUrd, inclusion of a 0 48HRSDEX 3-hour exposure to 1.5 pg/ml 33258 Hoechst dye (which in itself has no effect on cell viability) results in a 100-fold increase in light sensitivity. A comparison was made of the cloning efficiency of cultures grown for 48 hours in the presence or absence of dexamethasone and exposed to BrdUrd during the final 24 hours, treated with 33258 Hoechst, and irradiated. The results (fig. 7) indicate that there is no protection of the dexamethasone survivors against the effects of irradiation. If the GI effect of dexamethasone were reversible we would have expected to see decreased killing by the BrdUrd-Hoechst light regimen in the dexamethasone-treated cells. Since this was not the case we conclude that the G, effect is essentially irreversible. One possible complication in this type of exI I I I I 1 2 3 4 5 periment is that the presence of BrdUrd itself MINUTES IRRADIATION could cause suppression of the glucocorticoid Fig. 7 Effect of dexamethasone on photosensitivity of effect since BrdUrd has been shown to block Brd Urd substituted cplls. Parallel cultures were incuexpression of differentiated functions in some bated in the presence ( 0 )or absence (0) of M dexasystems including the induction of tyrosine methasone for 48 hours. BrdUrd (2 X lo-’ M) was added aminotransferase by glucocorticoids (Stellto each culture for the final 24 hours followed by three hours in the presence of 1.5 pg/ml 33258 Hoechst. Irradiawagen and Tomkins, ’71; Rutter et al., ’73). tion of cells was performed as in figure 5. Survivors are This was shown not to be the case here by expressed on the basis of unirradiated samples, normalizing for the toxicity of dexamethasone in the dexametha- measuring the viability of CEM-C7 cells when sone Rx’d culture. Survival of the dexamethasone-treated exposed to dexamethasone and BrdUrd indiculture was 7.3%. vidually and in combination; the toxicity of dexamethasone was not affected by the presence of BrdUrd (table 3).
k,
TABLE 3
DISCUSSION Effect of Brd Urd on dexamethasone toxicity Additions
None 48-hourdexamethasone 48-hour BrdUrd Dexamethasone + BrdUrd
% survivors
100.0 7.8 101 9.2
’
‘ Plating efficiency of untreated cells was used a s 100%.Actual plating efficiency was 65%. (Stetten et al., ’76). If the steroid effect is reversible, the arrested cells should subsequently form colonies when plated in the absence of steroid. The ability of 33258 Hoechst to potentiate the killing of BrdUrd substituted cells is shown in figure 6. A concentration of 2 x lo-’ M BrdUrd in a 24-hour incubation results in no
We have investigated the effects of dexamethasone, a potent synthetic glucocorticoid, on cell viability and cell cycle distribution of a steroid-sensitive clone of a human T lymphoblast cell line initially isolated from a patient with ALL (Foley e t al., ’65). The results obtained showed that the accumulation of cells with a DNA content characteristic of cells in the GI phase of the cell cycle was a steroid specific effect, related to the concentration of steroid used. Whether or not these cells are actually in G, is indeterminate and depends to some extent upon the definition of G,. It could be argued that these cells are in Go or have passed through the “A” state of Smith and Martin (‘73) to a terminally differentiated state of death. Being unable to discriminate among these possibilities we have defined all
276
HARMON, NORMAN, FOWLKES AND THOMPSON
cells with DNA contents characteristic of true GI cells as GI cells. It was previously shown (Norman and Thompson, '77) that when cells were continuously exposed in agarose gels to various glucocorticoidal and non-glucocorticoidal steroids, only those steroids with glucocorticoid activity (dexamethasone, prednisolone and cortisol) were effective in killing cells. Estradiol, dihydrotestosterone, and prednisone as well as the competitive antagonist to glucocorticoids, cortexolone, had no effect on cell viability. In those experiments it was not possible to assess the relative effects of those steroids which killed. However, when cells are exposed t o equal, near-saturating concentrations M) of prednisolone and dexamethasone for two days and then plated in the absence of steroid, the survival of prednisolone-treated cells was 80% while the survival of the dexamethasone-treated cells was only 10%(data not presented). This result is consistent with the lower affinity of prednisolone for receptors in rat thymocytes (Munck and Brinck-Johnsen, '68). The lethal response observed to various concentrations of dexamethasone was also consistent with the binding affinity of the steroid to receptor previously determined for this cell line (Norman and Thompson, '77). Both halfmaximal and maximal 48-hour effects corresponded to the equilibrium dissociation constant and saturation concentration respectively for the interaction between dexamethasone and its receptor. Thus on the basis of steroid specificity, saturability, and dose dependence we conclude that the killing effects observed are specific glucocorticoid responses. These results are analogous to the thymolytic effects of glucocorticoids in vivo (Claman, '72; Dougherty and White, '45) and with the effects of glucocorticoids on glucose transport (Rosen et al., '72) some amino acid transport (Makman et al., '68) and viability (Stuart and Ingram, '71) of rodent thymocytes and circulating T lymphocytes in vitro. The perturbation of the cell cycle by dexamethasone also meets the criteria of saturability and dose-dependence expected for a steroid-specific response. This effect is extremely interesting from both a clinical and biochemical point of view. Studies on clinical material from patients with ALL have shown that the labeling index for cells treated with 3H-thymidine in vitro began to fall approx-
imately 24 hours after treatment of patients with prednisolone (Ernst and Killman, '70; Lampkin et al., '71). This decrease preceded a fall in the mitotic index for the same cells and it was suggested that these results reflected a GI block induced by the steroid. The analysis of CEM-C7 cells by direct cell cycle measurements using flow-microfluorometric analysis of steroid-treated cells revealed that there is indeed a specific accumulation of cells in t h e G, phase of the cell cycle. The 24-hour delay between treatment and accumulation is consistent with the in vivo result and is nearly coincident with the onset of loss of cell viability. A reversible cell cycle specific mode of action for glucocorticoids would be important in planning chemotherapy which commonly includes glucocorticoid hormones (Goldin e t al., '71). The simultaneous use of cell cycle specific drugs acting at different parts of the cycle, a common practice in the treatment of ALL, might serve to decrease their combined effectiveness. In addition, a reversible GI block induced by glucocorticoids could be used to isolate genetic variants containing thermolabile steroid receptors, steroid resistant analogs to the "deathless" dibutyrl-CAMP resistant S49 variants of Lemaire and Coffino ('77), and steroid sensitive revertants of variant resistant lines. However, several lines of evidence suggest that the GI block is not reversible. The temporal relation of GI accumulation to cell death shows that GI accumulation is coincident with the onset of loss of cell viability. One would predict that if t h e GI block were reversible cells would pile up in GI and only after a period of time begin t o die. Indeed the reversible inhibition of cell growth by dibutyrl-CAMP is characterized by an almost total presence of cells in GI after 24 hours of exposure while cell viability remained a t 100% (Coffino et al., '75). Such is also the case for temperature sensitive mouse and Chinese hamster cell cycle mutants which are blocked in G I ; the cell cycle effects are observed to precede loss of viability (Talavera and Basilico, '77; Liskay, '74). In contrast, our observations on the distribution of cells in the cell cycle following the removal of dexamethasone (table 2) suggest the irreversibility of the GI accumulation; there were nearly as many cells in G, 12 hours after the removal of dexamethasone as there were when the dexamethasone was not removed.
G, ARREST AND KILLING OF CELLS BY DEXAMETHASONE
277
ACKNOWLEDGMENTS The inability of dexamethasone to protect cells from the BrdUrd-Hoechst-light regimen J. M. H. is a Fellow in cancer research supemployed also suggests that the GI accumula- ported by Grant DRG-116-F of the Damon tion is irreversible. Cells reversibly blocked in Runyon-Walter Winchell Cancer Fund. GI should not have incorporated significant LITERATURE CITED amounts of BrdUrd into their DNA and thus should have been less sensitive to the lethal Bourgeois, S., and R. F. Newby 1977 Diploid and haploid states of the glucocorticoid receptor gene of mouse effects of Hoechst-amplified BrdUrd-light lymphoid cell lines. Cell, 11: 423-430. killing. Claman, H. N. 1972 Corticosteroids and lymphoid cells. These experiments do not explain the mechN. Eng. J. Med., 287: 388-397. anism of the GI arrest. The accumulation of Clausen, 0.P.F., K. M. Gautvik and E. Haug 1978 Effects of cortisol, 17 P-estradiol and thyroliberin on prolactin cells in GI could be an effect of cell death or its and growth hormone production in cultured rat pituitary cause. However, i t is unlikely that the death tumour cells. J. Cell. Physiol., 94: 205-213. of each cell precedes its initial entry into GI. Coffino, P., J. W. Gray and G. M. Tomkins 1975 Cyclic AMP, a nonessential regulator of t h e cell cycle. Proc. The addition of colcemid to cultures exposed Natl. Acad. Sci. (U.S.A.), 72: 878-882. M dexamethasone for 24 hours showed to H. A,, and R. A. Tobey 1974 Cell cycle analysis in t h a t 20%of all cells are fixed in GI, a fraction Crissman, twenty minutes. Science, 184: 1297-1298. nearly identical to the fraction of nonviable Dougherty, T., and A. White 1945 Functional alterations in lymphoid tissue induced by adrenal corticoid secretion. cells. Since the GI accumulation is essentially Am. J. Anat., 77: 81-116. irreversible these 20%should be entirely nonP.,and S.-A. Killman 1970 Perturbation of generaviable and therefore can reasonably account Emst, tion cycle of human leukemic blast cells by cytostatic for the observed loss of cell viability. Nevtherapy in uiuo: Effect of cortico-steroids. Blood, 36: 689-696. ertheless, since viability has been measured by colony-forming ability, the presence at Foley, G. E., H. Lazarus, S. Farber, B. G. Uzman, B. A. Boone and R. E. McCarthy 1965 Continuous culture of human later times (table 2) of apparently cycling lymphoblasts from peripheral blood of a child with acute cells in greater numbers than colony-forming leukemia. Cancer, 18: 522-529 units indicates that not every cell which has Goldin, A., J. Sandberg, E. Henderson, J. Newman, E. Frei and J. Holland 1971 The chemotherapy of human and lost its colony-forming ability has been aranimal acute leukemia. Cancer Chemother. Rept., 55: rested in GI. Arrest in GI is lethal and irre309-507. versible, though a small fraction of cells are Harris, A. W. 1970 Differentiated functions expressed by cultured mouse lymphoma cells. Exp. Cell Res., 60: killed but not arrested after 60 hours. 341-353. The ability of glucocorticoids to induce a J. F., and A. White 1968 Effect of a single injection cell cycle specific response has been previous- Hofert, of cortisol on the incorporation of %thymidine and 3Hly demonstrated for the induction of the endeoxycytidine into lymphatic tissue DNA of adrenalectomized rats. Endocrinology, 82: 767-776. zyme tyrosine aminotransferase, which is K., and A. W. Harris 1970 Mouse myelomas and inducible in HTC cells through three-fourths Horibata, lymphomas in culture. Exp. Cell Res., 60: 61-77. of G, and all of S phase (Martin et al., ’69; Sel- Lampkin, B. C., T. Nagao and A. M. Mauer 1971 Synchronization and recruitment in acute leukemia. J. Clin. lers and Granner, ’74). Recently, Clausen et al. Invest., 50: 2204-2214. (‘78) have shown that glucocorticoids cause Lemaire, I., and P. Coffino 1977 Cyclic Amp-induced cytoaccumulation of GH3 cells in G,. More genlysis in S49 cells; selection of a n unresponsive “Deatherally, expression of many differentiated funcless” mutant. Cell, 11: 149-155. tions occurs following mitosis when cells have Liskay, R. M. 1974 A mammalian somatic “cell cycle” mutant defective in G,. J. Cell. Physiol., 84: 49-56. a DNA content characteristic of proliferating Makman, M. H., B. Dvorkin and A. White 1968 Influence of cells in GI. Therefore, the effects we have obcortisol on the utilization of nucleic acids and protein by lymphoid cells in uitro. J. Biol. Chem., 243: 1485-1497. served for dexamethasone on human T-lymM. H.,S. Nakagawa, B. Dvorkin and A. White phoblasts are consistent with models relating Makman, 1970 Inhibitory effects of cortisol and antibiotics on subgene-specific responses to steroid hormone strate entry and ribonucleic acid synthesis in rat thymoaction. cytes in uitro. J. Biol. Chem., 245: 2556-2563. D., G. M. Tomkins, and D. Granner 1969 Synthesis From the data presented we conclude that Martin, and induction of tyrosine aminotransferase in synboth cell death and GI accumulation of dexachronized heDatoma cells in culture. Proc. Natl. Acad. Sci. (U.S.A.),62: 248-255. methasone-treated cells are steroid specific responses. The G, accumulation is an irrever- Miller, M. H., J. I. Powell, S. 0. Sharrow and A. R. Schultz 1978 Rapid data collection, analysis and graphics for flow sible phenomenon, the relationship of which microfluorometry instrumentation. Rev. Sci. Instrum., to cell death remains to be established. 49: 78-83.
278
HARMON, NORMAN, FOWLKES AND THOMPSON
Munck, A,, and T. Brinck-Johnsen 1968 Specific and nonspecific physiochemical interactions of glucocorticoids and related steroids with r a t thymus cellsin uitro. J. Biol. Chem., 243: 5556-5565. Norman, M. R., and E. B. Thompson 1977 Characterization of a glucocorticoid-sensitivehuman lymphoid cell line. Cancer Res., 37: 3785-3791. Rosen, J. M., J. Fina, R. Milholland and F. Rosen 1972 Inhibitory effect of cortisol in uitro on 2-deoxyglucose uptake and RNA and protein metabolism in lymphosarcoma P1798. Cancer Res., 32: 350-355. Rutter, W. J., R. L. Pictet and P. W. Morris 1973 Toward molecular mechanisms of developmental processes. Annu. Rev. Biochem., 42: 601-616. Sellers, L., and D. Granner 1974 Regulation of tyrosine aminotransferase activity in two liver-derived permanent cell lines. J. Cell. Biol., 60: 337-345. Sibley, C. H., and G. M. Tomkins 1974 Mechanisms of steroid resistance. Cell, 2: 221-227. Smith, J. A., and L. Martin 1973 Do cells cycle? Proc. Natl. Acad. Sci. (U.S.A.), 70: 1263-1267. Steinkamp, J. A,, M. J. Fulwyler, S. R. Coulter, R. D. Hie-
bert, S. L. Homey and P. F. Mullaney 1973 A new multiparameter separator for microscopic particles and biological cells. Rev. Sci. Instrum., 44: 1301-1310. Stellwagen, R. H., and G. M. Tomkins 1971 Preferential inhibition by 5-bromodeoxyuridine of the synthesis of tyrosine aminotransferase in hepatoma cell cultures. J. Mol. Biol., 56: 167-182. Stetten, G., S. A. Latt and R. L. Davidson 1976 33258 Hoechst enhancement of the photosensitivity of bromodeoxyuridine-substituted cells. Somatic Cell Genetics, 2: 285-290. Stuart, J. T., and M. Ingram 1971 The effect of cortisol on viability and glucose uptake in rat thymocytes in uitro. Proc. Soc. Exp. Biol. Med., 136: 1146-1150. Talavera, A., and C. Basilico 1977 Temperature sensitive mutants of BHK cells affected in cell cycle progression. J. Cell. Physiol., 92: 425-436. Yamamoto, K. R., M. R. Stampfer and G. M. Tomkins 1974 Receptors from glucocorticoid-sensitivelymphoma cells and two classes of insensitive clones: Physical and DNA binding properties. Proc. Natl. Acad. Sci. (U.S.A.), 21: 3901-3905.
ABSTRACT
Growth of a human leukemic T-cell line (CEM C7) in M dexamethasone results in inhibition of growth and rapid loss of cell viability after a delay of approximately 18 to 24 hours. Analysis of dexamethasonetreated cells by flow-microfluorometry showed that they were arrested in the GI phase of t h e cell cycle. Loss of cell viability began at the same time as G, accumulation was first detectable, and 20% of all cells were found to be blocked in G, at this time suggesting that loss of viability and GI arrest were coincident events. Half-maximal and maximal effects on both viability and GI arrest after 48 hours in steroid were nearly identical with respect to steroid concentration and corresponded to half-maximal and full occupancy of glucocorticoid specific receptor by hormone, consistent with a glucocorticoid receptor mediated mechanism for both phenomena. Most non-viable cells were arrested in G1,and accumulation of cells in G1 was irreversible; removal of steroid in the presence of colcemid did not result in a decreased fraction of G, cells. Furthermore, dexamethasone treatment did not protect cells against the effects of 33258 Hoechstamplified killing of bromodeoxyuridine substituted cells exposed to light. These results show t h a t dexamethasone arrests these leukemic cells in GI and strongly suggest t h a t dexamethasone-treated cells are killed upon entry into GI.
'
The multiple effects of glucocorticoid hormones on lymphoid tissue have been widely studied and documented. Inhibitory effects on a number of cellular processes have been described including those on transport of glucose and some amino acids (Rosen et al., '72; Makman et al., '681, on incorporation of precursors into RNA and DNA (Makman et al., '70; Hofert and White, '68),and ultimately on cell growth. Glucocorticoids cause lysis of rodent thymus cells and thymus-derived cell lines both in vivo (Dougherty and White, '45; Claman, '72), and in vitro (Harris, '70; Bourgeois and Newby, '77). Lysis of untransformed thymus cells has been difficult to study in vitro due to the relatively short period of time during which t h e tissue or dissociated cells can be maintained in a viable state. Consequently, several rodent lymphoid tissue cult u r e lines have been developed to facilitate such in vitro studies (Harris, '70; Horibata J. CELL. PHYSIOL. (1979) 98: 267-278.
and Harris, '70). We have recently characterized a clonal human lymphoblastoid cell line (CEM-C7) derived from the CCRF-CEM line, which was originally established from a patient with acute lymphoblastic leukemia. The CEM-C7 line is sensitive to glucocorticoid hormones and is the first human line in which glucocorticoid receptor occupancy has been correlated with the growth inhibitory effects of the steroid (Norman and Thompson, '77). In this sense the CEM cell line appears analogous to t h e S49 mouse lymphoma line, from which several classes of steroid resistant mutants have already been isolated (Sibley and Tomkins, '74; Yamamoto et al., '74). The mechanism of the inhibitory effects of Received Apr. I, '78.Accepted Aug. 30, '78. ' Permanent address: Department of Chemical Pathology, Kings College Hospital, Landon, England. Present address: Laboratory of Microbial Immunity, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, Maryland 20014.
267
260
HARMON, NORMAN, FOWLKES AND THOMPSON
glucocorticoids on cell growth is still unknown. However, the possibility that glucocorticoids produce their growth inhibitory effects on lymphoid cells by blocking them in the G, phase of the cell cycle was suggested by studies in which patients with acute lymphoblastic leukemia (ALL) were treated with glucocorticoids. Lymphoblasts from these patients were found to have a decreased 3H-thymidine labeling index several hours before a decrease in mitotic index was observed (Ernst and Killman, '70; Lampkin et al., '71). We felt that it was important to determine whether steroids indeed block leukemic cells a t a specific point in the cell cycle. If so, it would have important implications for mechanisms of cell killing. The reversibility, or irreversibility of the block would have important pharmacologic implications, inasmuch as two reversible cycle-active drugs used in combination during chemotherapy might interfere with each other. If the effect were reversible, it would also provide unique advantages for isolation of conditional steroid resistant variants. We therefore have used the CEM-C7 cell line to study the lethal effects of glucocorticoids on human T cells. In particular we have found that steroid treatment of these cells results in an irreversible accumulation of cells in G,.
loid fibroblasts (CRL. 1221, American Type Culture Collection) was seeded onto the desired number of 60-mm plates. Just prior to the plating of cells a 2.5-ml separator layer of 10% FCS containing 0.25% RMPI 1640 agarose (Type 11, Sigma Chemical Co.) maintained a t 42°C was pipetted onto each plate and allowed to gel. Immediately thereafter, 1 volume of a cell suspension of appropriate density was mixed with 4.5 volumes of RPMI 1640 a t twice isotonic concentration, containing 20% FCS a t 37°C. To this suspension 4.5 volumes of 0.5% agarose a t 42°C were ,quickly added and 2.5 ml of this mixture were rapidly pipetted over each separator layer and allowed to gel a t room temperature. Plates were incubated for two to three weeks, stained with neutral red and the number of colonies determined. All colonies containing more than 100 cells were counted. In separate experiments efficiency of colony formation varied between 40 and 60%.Within a single experiment, however, cloning efficiency of multiple control cultures was constant.
+
Flow-microfluorometry
Cell cycle analyses were obtained using the fluorochrome, mithramycin (Pfizer), the cellular binding of which has been shown to be stoichiometrically related to DNA content MATERIALS AND METHODS (Crissman and Tobey, '74). Cultured cells were washed two times with phosphate buffered Cell growth and culture saline (PBS: KC1, 0.2 glliter; KH2P04,0.2 g/ The clonal cell line CEM-C7 was cloned from liter; NaC1, 8.0 glliter; and NaZHPO4,2.16 g/ the cell line CCRF-CEM as previously de- liter) and fixed in 70%ethanol. The fixed cells scribed (Norman and Thompson, '77). Cells were resuspended in 3-5 ml of staining soluwere grown in RPMI 1640 (NIH Media Unit) tion (0.85%NaCl solution containing 15 mM containing 10% heat inactivated fetal calf MgClz and 100 pg/ml mithramycin) for a t serum (North American Biologicals).Cultures least 20 minutes a t room temperature. Cells were maintained a t densities of 5 x lo4- 2 x were filtered through a 62-p nylon mesh to re106per ml a t 37°C in a humidified atmosphere move any clumps. The concentration of cells of 95%air, 5%COz,as stationary suspensions. in staining solution was adjusted to approxUnder these conditions repeated determina- imately 4-6 X 105/mlto minimize coincidence tions showed a population doubling time of 18 errors during analysis. to 20 hours. Cell number was determined usThe DNA content of individual cells was aning a Model B Coulter counter (Coulter Elec- alyzed by measuring the mithramycin stained tronics). The lower threshold was set so that cells using the Los Alamos Scientific Laborawhen cells were treated with dexamethasone, tory (LASL) Multiparameter Analyzer and pyknotic cells would not be excluded. Cell Sorter. The design and operation of this instrument has been described by Steinkamp Determination of cell viability et al. ('73). A 457-nm argon ion laser line was Cell viability was assayed by determining used for excitation, and total fluorescent the colony-forming efficiency of single cells emission above 495 nm was measured. Histoplated in semisolid medium. Several days grams of the DNA spectra were produced by a before each experiment a layer of human dip- DEC PDP 11/40 Computer (Maynard, Massa-
269
G, ARREST AND KILLING OF CELLS BY DEXAMETHASONE
chusetts) with an RT 11 Operating System. The data analysis, storage, and graphics used were as described by Miller e t al. (’78). The determination of DNA content on individual cells yields a distinctive distribution spectrum (cf. fig. 2). The spectrum shows two peaks; the first, that of cells with a G, content of DNA, the second t h a t of cells with double t h e G, content; that is cells in G Por M phase. The interval between the two peaks corresponds to cells in S phase. This distribution was then analyzed according to the method of Miller et al. (‘78) to derive the percentage of cells in GI, S, and Gz M. Compartment boundaries were selected for integration using control samples and were maintained for all subsequent analyses. One hundred thousand cells were analyzed t o generate each histogram. Calibration of instrument for linearity and coefficient of variation using standard particles (Particle Technology, Inc., Los Alamos, New Mexico) was always done prior to analyzing samples. The coefficient of variation of these particles was maintained between 4.04.5%throughout these studies. 33258 Hoechst-amplified BrdUrd-light killing Cells were photosensitized essentially by t h e method of Stetten et al. (‘76). Cultures were grown in the presence of BrdUrd (5-bromodeoxyuridine, Sigma) for 24 hours. Freshly prepared 33258 Hoechst (Riedel-DE Haen Ag Seelze, Hannover) was added to the culture to give a final concentration of 1.5pg/ml and the cells returned to the incubator for a n additional three hours. Cells to be irradiated were transferred to fresh medium and 2.5 ml of cell suspension placed into 60-mm covered tissue culture dishes. Dishes were exposed from above to two Sylvania Cool White fluorescent lamps (F15 T 8 - CW, Sylvania) at a distance of 6 cm for up to five minutes. Irradiated cells were plated at 250, lo3, and lo4cells/plate in quadruplicate and plates with 20 to 200 colonies per plate were counted as described above. Fractional survival was determined as the percent colony formation of the treated cells divided by t h e percent colony formation of a n untreated control.
I
+
,
t
RESULTS
Effects of dexamethasone on CEM-C7 cells Addition of dexamethasone to logarith-
1
2
3
4
I 1
I 2
I 3
4
0.
0.0.
I
DAYS Fig. 1 Inhibition of cell growth and viability by dexaM (0- O ) , M methasone. Cells were grown in (0-U), M (A-A) or no (0-0)dexamethasone and periodically assayed for cell number (upper panel) or cell viability (lower panel) as described in MATERIALS AND METHODS.
mically growing cultures of CEM-C7 results in inhibition of cell proliferation after a delay of about 24 hours (fig. 1). This inhibition is accompanied by a progressive loss of cell viability as measured by the loss of plating efficiency in agarose gels (fig. 1). The assumption that the ability to form clones measures
270
HARMON, NORMAN, FOWLKES AND THOMPSON
S
14.5r
- 17.0
G,+M
a
10.9
12.8
I
N
X
I
CHANNEL NUMBER Fig. 2 Influence of dexamethasone on cell cycle distribution. Cells were grown in the presence (b) or absence (a) of dexamethasone for 48 hours and then analyzed by flow-microfluorometryas described in MATERIALS AND METHODS. Abscissa: Channel No. is linear function of DNA/cell. Ordinate: A total of lo5 cells were counted for each histogram and scale difference in the two panels reflects the fact that display of data is normalized to give a uniform maximum peak height.
viability derives from the fact that glucocorticoids cause lysis of CEM-C7 cells in mass culture (Norman and Thompson, '77) and that direct observation of the cells cloned in agarose after steroid treatment showed no intact cells or abortive colonies but only cell debris vs surviving colonies. To ascertain whether these effects were accompanied by cell cycle specific inhibition, the cell cycle distributions of dexamethasone-treated and untreated cultures were studied by flow-microfluorometry (FMF). The results (fig. 2) showed that after 48 hours of exposure to M dexamethasone a far greater proportion of cells had a DNA content characteristic of G, cells than did those of an untreated culture. No clumping of cells was observed in either the steroid treated or untreated cultures. Therefore differential filtration through the 62-pnylon mesh during preparation of cells for FMF cannot explain this result.
Relationship of killing to cell cycle perturbation To determine if the shift in the cell cycle distribution and loss of cell viability were related, the time course of both loss of viability and alteration in cell cycle distribution was observed. Cells were treated with M dexamethasone and a t various times thereafter removed from the steroid, plated for viability, and analyzed by FMF. The results (fig. 3) show that cells begin to lose viability 24 hours after exposure to the steroid. Following this initial delay, loss of viability proceeds rapidly, with a rate of loss of about 50%per six hours. A shift in cell cycle distribution is detectable 18-24 hours after exposure to steroid and cells continue to accumulate in G , thereafter, suggesting a possible temporal relationship between loss of viability and cell cycle perturbation. It was not possible to study the cell cycle distribution beyond 60 hours of steroid treatment
271
G I ARREST AND KILLING OF CELLS BY DEXAMETHASONE
due to cell lysis and accumulation of cellular debris which produced FMF patterns impossible to interpret. One of the characteristics expected for a steroid specific response is that the magnitude of the response be related to the fraction of occupied steroid receptor molecules and therefore be saturable. To determine if loss of cell viability and altered cell cycle distribution met this criterion the dose responsiveness of both loss of viability and cell cycle perturbation was determined. Cultures of CEM-C7 were incubated with various concentrations of dexamethasone for 48 hours and the cells assayed for both viability and cell cycle distribution. The results (fig. 4) show that both 50% loss of viability and 50%of the maximal effect on the cell cycle are obtained in the presence of 2-5 X lo-’ M dexamethasone. This corresponds closely to the previously measured & for the steroid receptor in these cells of 1.4 x lo-’ M (Norman and Thompson, ’77).The maximal effects on both viability and Gl shift ocM dexamethacur a t approximately 5 X sone which is in good agreement with the concentration a t which receptors are saturated (Norman and Thompson, ’77).
Nature o f the cell cycle effect The shift in the cell cycle distribution of dexamethasone-treated cells to one apparently enriched in GI cells could have been the result of an actual accumulation of cells in the G1 phase of the cell cycle or of differential lysis of cells in the other phases of the cell cycle thereby establishing a new distribution. If the latter possibility were the case then cells which are in G, should be able to enter S; whereas if cells are actually accumulating in G1 they should be unable to do so. To distinguish between these possibilities, a population M dexamethasone of cells treated with for 24 hours was exposed to the mitotic blocker colcemid (0.5 pg/ml) for an additional 12 hours in the continued presence of dexamethasone. When the cells were then analyzed by FMF the results showed that while some decrease in the fraction of cells in G1 occurred during the colcemid treatment, approximately 60% of the cells which were in GI when the colcemid was added were still there 12 hours later (table 1comparing lines 3 and 4). Phase contrast microscopy of the culture a t this time revealed no significant cell lysis. Since colcemid treatment of control cells re-
sulted in an almost complete depletion of cells from GI (table 1,line 2), we conclude that the effect of dexamethasone on the cell cycle is to cause an accumulation of cells in G,. In addiM tion, the increase in the fraction of G, cells during the combined treatment indicates that those cells not arrested in G, continue to cycle (table 1,lines 3 and 4). The accumulation of cells in G, could result from two mechanisms. Either cells which have already lost colony forming ability manage to “coast” through the cell cycle until they reach GI or cells lose this ability when they actually enter G,. The former possibility would predict
+
loo
r
I
CELLS IN GI
12
0.01
24
3 6 4 8 6 0
k 12
24
36
48
60
TIME (hrs) Fig. 3 Time course of loss of viability and G Iaccumulation. Cells were grown in M dexamethasone for various lengths of time and then assayed for cell cycle distribution (upper panel) and viability flower panel) as de-
scribed in MATERIALS AND METHODS.
272
HARMON, NORMAN, FOWLKES AND THOMPSON
100'
80
2
>a l-
a
240 UI n
20
0
10-8
I
I 5~10-~
2x10-8
I
1o
I 2x10-7
-~
I 5x10-7
10-6'-
DEXAMETHASONE (M) Fig. 4 Effect of dexamethasone concentration on viability and cell cycle distribution. Cells were grown for 48 hours in various concentrations of dexamethasone and then analyzed for cell cycle distribution ( A ) and
viability (0).
TABLE 1
Early effect of dezamethasone on G Iaccumulation Distribution of cells (%)
Treatment Dexamethasone
Colcemid
-
-
-
12hours
24 hours 36 hours
12hours
-
G,
S
G,+M
46 6 49 31
26 21 13 18
28 14
38 51
M dexamethaaone. A culture of logarithmically growing CEM-C7 cells was divided in half. One-half was incubated with Twenty-four hours later a sample of each was analyzed by FMF (lines 1and 3). The control culture was again divided and colcemid was added to one-half. Colcemid was also added to the culture containing dexamethasone. Twelve hours later, each culture was again analyzed by FMF (lines 2 and 4). The control culture gave t h e same distribution as a t 24 hours ' Determined a s described In MATERIALS AND METHODS. One-half microgramfmilliliter for terminal 1 2 hours.
that loss of viability should precede cell cycle arrest, while the latter would predict that loss of cell viability should either follow or be coincident with cell cycle arrest. Comparison of the curves in the upper and lower panels of figure 3 shows that the onset of G1accumulation and initial loss of viability correspond closely in time. However, the quantity of cells in GI shown in figure 3 cannot be used for comparison since an altered cell cycle distribution becomes apparent only after cells distal to a cell cycle specific block have passed out of the
phase of the cycle in which the block occurs. Thus, early or mid-G, arrest would not be reflected in an altered FMF distribution for several hours after the first cells were arrested. A more direct assessment of the fraction of cells arrested in G, at a specific time may be obtained by adding colcemid to cells at the same time as steroid is removed from the medium. Subsequent growth of the cells will allow unaffected cells to exit G1,leaving only arrested cells to be detected by FMF analysis. Such an experiment was performed on cells grown in
273
G I ARREST AND KILLING OF CELLS BY DEXAMETHASONE
20
I
a
15
16.7
cv
c'!
z
z
X
X
2
5
10
J
-1
0
u
11.1
w
w
5
8.9
8
6.7
6
cv
c'!
z
2
X
y
I
X
4.4
Y 4 -I
J
w
W 0
0
2.2
2
4 . I
200
I
I
400
I
I
200 CHANNEL NUMBER
I
I
400
I
600
Fig. 5 G , arrested cells after 24 hours in dexamethasone. Cells were grown in the absence (a, b) or presence (c, d) of dexamethasone for 24 hours. Dexamethasone was then removed and 0.5 pg/ml colcemid was added to one untreated (b) and one steroid treated (d) culture. Growth was continued for 12 hours a t which time cells were analyzed by FMF as described in MATERIALS AND METHODS.
M dexamethasone for 24 hours. Twelve hours of colcemid treatment obliterates the characteristic G , peak of control cells (fig. 5b). However, colcemid treatment of cells exposed M dexamethasone for 24 hours leaves a to
significant fraction of cells in G , (fig. 5d). This fraction, about 20% of the entire population, therefore seems to be arrested in G I . This result is in good agreement with that of the experiment of figure 3, which showed a compara-
274
HARMON, NORMAN, FOWLKES AND THOMPSON
ble loss of viability after 24-hour treatment with dexamethasone. These results clearly show that loss of cell viability does not precede cell cycle arrest. The possibility that the GI accumulation is the result of cells which have been killed elsewhere but have managed to reach G , is thus excluded.
o CONTROL 0
HOECHST
A BrdUrd
A BrdUrd
0.001
I
I 1
1
I
2
3
I 4
+ HOECHST
I 5
MINUTES IRRADIATION
Fig. 6 Photosensitization of BrdUrd substituted cells by 33258 Hoechst. Parallel cultures (4) were grown in the presence or absence of 2 X lo-' M BrdUrd for 24 hours. 33258 Hoechst (1.5 pg/ml) was added to one of the BrdUrd substituted and one of the unsubstituted cultures and all four cultures were incubated for an additional three hours. Cells were resuspended in fresh medium and irradiated and assayed for viability as described in MATERIALS O ) , 3 hours AND METHODS. (0-01,no addition; (0in 33258 Hoechst; (A-A), 24 hours in BrdUrd; (A-A), 24 hours BrdUrd, 3 hours 33258 Hoechst.
Irreversibility of cell cycle effect To determine if cell cycle arrest by dexamethasone was a reversible effect two parallel M dexamethacultures were incubated in sone for 48 hours, after which the cells were resuspended in fresh medium. One of the culM dexatures was resupplemented with methasone and both cultures were incubated for an additional 12 hours. Cells from both cultures and a control, untreated culture were then analyzed by FMF. The results of this experiment are presented in table 2. The absence of dexamethasone during the final 12 hours of incubation resulted in no significant reversal of the G, accumulation. Although such a reversal may occur in the small fraction of surviving cells (figs. 1, 3), this would be difficult to detect since this fraction (-7%) represents only a small proportion of the total number of cells analyzed. However, 95%,not 75% of the cells are incapable of forming colonies after 60 hours of dexamethasone (fig. 3, table 2). Therefore there are many cells outside of GI which are not capable of forming colonies. The inability of these cycling cells to form colonies when steroid is removed most likely indicates that these cells have sustained insufficient damage in G I to prevent a limited amount of cycling but are sufficiently damaged to prevent the degree of proliferation necessary to clone in semisolid medium. This view is supported by microscopic examination of dexamethasone-treated cells at this time which shows virtually all of them to be severely deformed (Norman and Thompson, '77). To determine whether the small steroid surviving fraction is composed of cells reversibly arrested in GI we employed a second approach. Cells blocked in GI should incorporate sharply reduced amounts of BrdUrd into their DNA. Such cells should therefore be less sensitive to the lethal effects of incubation in 33258 Hoechst dye followed by exposure to light
TABLE 2
Reversibility of G Iaccumulation Distribution of cells (%)
n m e of treatment
I
'
0-48
48-60
G,
S
G,+M
-
-
Dexamethasone Dexamethasone
Dexamethasone
55 76 15
22 12 14
23 12 11
-
All determinations were made at 60 hours.
G, ARREST AND KILLING OF CELLS BY DEXAMETHASONE
275
loss of cell viability. Five minutes of irradiation of such BrdUrd-treated cells results in a 70%loss of viability. However, with cells ideno CONTROL tically incubated with BrdUrd, inclusion of a 0 48HRSDEX 3-hour exposure to 1.5 pg/ml 33258 Hoechst dye (which in itself has no effect on cell viability) results in a 100-fold increase in light sensitivity. A comparison was made of the cloning efficiency of cultures grown for 48 hours in the presence or absence of dexamethasone and exposed to BrdUrd during the final 24 hours, treated with 33258 Hoechst, and irradiated. The results (fig. 7) indicate that there is no protection of the dexamethasone survivors against the effects of irradiation. If the GI effect of dexamethasone were reversible we would have expected to see decreased killing by the BrdUrd-Hoechst light regimen in the dexamethasone-treated cells. Since this was not the case we conclude that the G, effect is essentially irreversible. One possible complication in this type of exI I I I I 1 2 3 4 5 periment is that the presence of BrdUrd itself MINUTES IRRADIATION could cause suppression of the glucocorticoid Fig. 7 Effect of dexamethasone on photosensitivity of effect since BrdUrd has been shown to block Brd Urd substituted cplls. Parallel cultures were incuexpression of differentiated functions in some bated in the presence ( 0 )or absence (0) of M dexasystems including the induction of tyrosine methasone for 48 hours. BrdUrd (2 X lo-’ M) was added aminotransferase by glucocorticoids (Stellto each culture for the final 24 hours followed by three hours in the presence of 1.5 pg/ml 33258 Hoechst. Irradiawagen and Tomkins, ’71; Rutter et al., ’73). tion of cells was performed as in figure 5. Survivors are This was shown not to be the case here by expressed on the basis of unirradiated samples, normalizing for the toxicity of dexamethasone in the dexametha- measuring the viability of CEM-C7 cells when sone Rx’d culture. Survival of the dexamethasone-treated exposed to dexamethasone and BrdUrd indiculture was 7.3%. vidually and in combination; the toxicity of dexamethasone was not affected by the presence of BrdUrd (table 3).
k,
TABLE 3
DISCUSSION Effect of Brd Urd on dexamethasone toxicity Additions
None 48-hourdexamethasone 48-hour BrdUrd Dexamethasone + BrdUrd
% survivors
100.0 7.8 101 9.2
’
‘ Plating efficiency of untreated cells was used a s 100%.Actual plating efficiency was 65%. (Stetten et al., ’76). If the steroid effect is reversible, the arrested cells should subsequently form colonies when plated in the absence of steroid. The ability of 33258 Hoechst to potentiate the killing of BrdUrd substituted cells is shown in figure 6. A concentration of 2 x lo-’ M BrdUrd in a 24-hour incubation results in no
We have investigated the effects of dexamethasone, a potent synthetic glucocorticoid, on cell viability and cell cycle distribution of a steroid-sensitive clone of a human T lymphoblast cell line initially isolated from a patient with ALL (Foley e t al., ’65). The results obtained showed that the accumulation of cells with a DNA content characteristic of cells in the GI phase of the cell cycle was a steroid specific effect, related to the concentration of steroid used. Whether or not these cells are actually in G, is indeterminate and depends to some extent upon the definition of G,. It could be argued that these cells are in Go or have passed through the “A” state of Smith and Martin (‘73) to a terminally differentiated state of death. Being unable to discriminate among these possibilities we have defined all
276
HARMON, NORMAN, FOWLKES AND THOMPSON
cells with DNA contents characteristic of true GI cells as GI cells. It was previously shown (Norman and Thompson, '77) that when cells were continuously exposed in agarose gels to various glucocorticoidal and non-glucocorticoidal steroids, only those steroids with glucocorticoid activity (dexamethasone, prednisolone and cortisol) were effective in killing cells. Estradiol, dihydrotestosterone, and prednisone as well as the competitive antagonist to glucocorticoids, cortexolone, had no effect on cell viability. In those experiments it was not possible to assess the relative effects of those steroids which killed. However, when cells are exposed t o equal, near-saturating concentrations M) of prednisolone and dexamethasone for two days and then plated in the absence of steroid, the survival of prednisolone-treated cells was 80% while the survival of the dexamethasone-treated cells was only 10%(data not presented). This result is consistent with the lower affinity of prednisolone for receptors in rat thymocytes (Munck and Brinck-Johnsen, '68). The lethal response observed to various concentrations of dexamethasone was also consistent with the binding affinity of the steroid to receptor previously determined for this cell line (Norman and Thompson, '77). Both halfmaximal and maximal 48-hour effects corresponded to the equilibrium dissociation constant and saturation concentration respectively for the interaction between dexamethasone and its receptor. Thus on the basis of steroid specificity, saturability, and dose dependence we conclude that the killing effects observed are specific glucocorticoid responses. These results are analogous to the thymolytic effects of glucocorticoids in vivo (Claman, '72; Dougherty and White, '45) and with the effects of glucocorticoids on glucose transport (Rosen et al., '72) some amino acid transport (Makman et al., '68) and viability (Stuart and Ingram, '71) of rodent thymocytes and circulating T lymphocytes in vitro. The perturbation of the cell cycle by dexamethasone also meets the criteria of saturability and dose-dependence expected for a steroid-specific response. This effect is extremely interesting from both a clinical and biochemical point of view. Studies on clinical material from patients with ALL have shown that the labeling index for cells treated with 3H-thymidine in vitro began to fall approx-
imately 24 hours after treatment of patients with prednisolone (Ernst and Killman, '70; Lampkin et al., '71). This decrease preceded a fall in the mitotic index for the same cells and it was suggested that these results reflected a GI block induced by the steroid. The analysis of CEM-C7 cells by direct cell cycle measurements using flow-microfluorometric analysis of steroid-treated cells revealed that there is indeed a specific accumulation of cells in t h e G, phase of the cell cycle. The 24-hour delay between treatment and accumulation is consistent with the in vivo result and is nearly coincident with the onset of loss of cell viability. A reversible cell cycle specific mode of action for glucocorticoids would be important in planning chemotherapy which commonly includes glucocorticoid hormones (Goldin e t al., '71). The simultaneous use of cell cycle specific drugs acting at different parts of the cycle, a common practice in the treatment of ALL, might serve to decrease their combined effectiveness. In addition, a reversible GI block induced by glucocorticoids could be used to isolate genetic variants containing thermolabile steroid receptors, steroid resistant analogs to the "deathless" dibutyrl-CAMP resistant S49 variants of Lemaire and Coffino ('77), and steroid sensitive revertants of variant resistant lines. However, several lines of evidence suggest that the GI block is not reversible. The temporal relation of GI accumulation to cell death shows that GI accumulation is coincident with the onset of loss of cell viability. One would predict that if t h e GI block were reversible cells would pile up in GI and only after a period of time begin t o die. Indeed the reversible inhibition of cell growth by dibutyrl-CAMP is characterized by an almost total presence of cells in GI after 24 hours of exposure while cell viability remained a t 100% (Coffino et al., '75). Such is also the case for temperature sensitive mouse and Chinese hamster cell cycle mutants which are blocked in G I ; the cell cycle effects are observed to precede loss of viability (Talavera and Basilico, '77; Liskay, '74). In contrast, our observations on the distribution of cells in the cell cycle following the removal of dexamethasone (table 2) suggest the irreversibility of the GI accumulation; there were nearly as many cells in G, 12 hours after the removal of dexamethasone as there were when the dexamethasone was not removed.
G, ARREST AND KILLING OF CELLS BY DEXAMETHASONE
277
ACKNOWLEDGMENTS The inability of dexamethasone to protect cells from the BrdUrd-Hoechst-light regimen J. M. H. is a Fellow in cancer research supemployed also suggests that the GI accumula- ported by Grant DRG-116-F of the Damon tion is irreversible. Cells reversibly blocked in Runyon-Walter Winchell Cancer Fund. GI should not have incorporated significant LITERATURE CITED amounts of BrdUrd into their DNA and thus should have been less sensitive to the lethal Bourgeois, S., and R. F. Newby 1977 Diploid and haploid states of the glucocorticoid receptor gene of mouse effects of Hoechst-amplified BrdUrd-light lymphoid cell lines. Cell, 11: 423-430. killing. Claman, H. N. 1972 Corticosteroids and lymphoid cells. These experiments do not explain the mechN. Eng. J. Med., 287: 388-397. anism of the GI arrest. The accumulation of Clausen, 0.P.F., K. M. Gautvik and E. Haug 1978 Effects of cortisol, 17 P-estradiol and thyroliberin on prolactin cells in GI could be an effect of cell death or its and growth hormone production in cultured rat pituitary cause. However, i t is unlikely that the death tumour cells. J. Cell. Physiol., 94: 205-213. of each cell precedes its initial entry into GI. Coffino, P., J. W. Gray and G. M. Tomkins 1975 Cyclic AMP, a nonessential regulator of t h e cell cycle. Proc. The addition of colcemid to cultures exposed Natl. Acad. Sci. (U.S.A.), 72: 878-882. M dexamethasone for 24 hours showed to H. A,, and R. A. Tobey 1974 Cell cycle analysis in t h a t 20%of all cells are fixed in GI, a fraction Crissman, twenty minutes. Science, 184: 1297-1298. nearly identical to the fraction of nonviable Dougherty, T., and A. White 1945 Functional alterations in lymphoid tissue induced by adrenal corticoid secretion. cells. Since the GI accumulation is essentially Am. J. Anat., 77: 81-116. irreversible these 20%should be entirely nonP.,and S.-A. Killman 1970 Perturbation of generaviable and therefore can reasonably account Emst, tion cycle of human leukemic blast cells by cytostatic for the observed loss of cell viability. Nevtherapy in uiuo: Effect of cortico-steroids. Blood, 36: 689-696. ertheless, since viability has been measured by colony-forming ability, the presence at Foley, G. E., H. Lazarus, S. Farber, B. G. Uzman, B. A. Boone and R. E. McCarthy 1965 Continuous culture of human later times (table 2) of apparently cycling lymphoblasts from peripheral blood of a child with acute cells in greater numbers than colony-forming leukemia. Cancer, 18: 522-529 units indicates that not every cell which has Goldin, A., J. Sandberg, E. Henderson, J. Newman, E. Frei and J. Holland 1971 The chemotherapy of human and lost its colony-forming ability has been aranimal acute leukemia. Cancer Chemother. Rept., 55: rested in GI. Arrest in GI is lethal and irre309-507. versible, though a small fraction of cells are Harris, A. W. 1970 Differentiated functions expressed by cultured mouse lymphoma cells. Exp. Cell Res., 60: killed but not arrested after 60 hours. 341-353. The ability of glucocorticoids to induce a J. F., and A. White 1968 Effect of a single injection cell cycle specific response has been previous- Hofert, of cortisol on the incorporation of %thymidine and 3Hly demonstrated for the induction of the endeoxycytidine into lymphatic tissue DNA of adrenalectomized rats. Endocrinology, 82: 767-776. zyme tyrosine aminotransferase, which is K., and A. W. Harris 1970 Mouse myelomas and inducible in HTC cells through three-fourths Horibata, lymphomas in culture. Exp. Cell Res., 60: 61-77. of G, and all of S phase (Martin et al., ’69; Sel- Lampkin, B. C., T. Nagao and A. M. Mauer 1971 Synchronization and recruitment in acute leukemia. J. Clin. lers and Granner, ’74). Recently, Clausen et al. Invest., 50: 2204-2214. (‘78) have shown that glucocorticoids cause Lemaire, I., and P. Coffino 1977 Cyclic Amp-induced cytoaccumulation of GH3 cells in G,. More genlysis in S49 cells; selection of a n unresponsive “Deatherally, expression of many differentiated funcless” mutant. Cell, 11: 149-155. tions occurs following mitosis when cells have Liskay, R. M. 1974 A mammalian somatic “cell cycle” mutant defective in G,. J. Cell. Physiol., 84: 49-56. a DNA content characteristic of proliferating Makman, M. H., B. Dvorkin and A. White 1968 Influence of cells in GI. Therefore, the effects we have obcortisol on the utilization of nucleic acids and protein by lymphoid cells in uitro. J. Biol. Chem., 243: 1485-1497. served for dexamethasone on human T-lymM. H.,S. Nakagawa, B. Dvorkin and A. White phoblasts are consistent with models relating Makman, 1970 Inhibitory effects of cortisol and antibiotics on subgene-specific responses to steroid hormone strate entry and ribonucleic acid synthesis in rat thymoaction. cytes in uitro. J. Biol. Chem., 245: 2556-2563. D., G. M. Tomkins, and D. Granner 1969 Synthesis From the data presented we conclude that Martin, and induction of tyrosine aminotransferase in synboth cell death and GI accumulation of dexachronized heDatoma cells in culture. Proc. Natl. Acad. Sci. (U.S.A.),62: 248-255. methasone-treated cells are steroid specific responses. The G, accumulation is an irrever- Miller, M. H., J. I. Powell, S. 0. Sharrow and A. R. Schultz 1978 Rapid data collection, analysis and graphics for flow sible phenomenon, the relationship of which microfluorometry instrumentation. Rev. Sci. Instrum., to cell death remains to be established. 49: 78-83.
278
HARMON, NORMAN, FOWLKES AND THOMPSON
Munck, A,, and T. Brinck-Johnsen 1968 Specific and nonspecific physiochemical interactions of glucocorticoids and related steroids with r a t thymus cellsin uitro. J. Biol. Chem., 243: 5556-5565. Norman, M. R., and E. B. Thompson 1977 Characterization of a glucocorticoid-sensitivehuman lymphoid cell line. Cancer Res., 37: 3785-3791. Rosen, J. M., J. Fina, R. Milholland and F. Rosen 1972 Inhibitory effect of cortisol in uitro on 2-deoxyglucose uptake and RNA and protein metabolism in lymphosarcoma P1798. Cancer Res., 32: 350-355. Rutter, W. J., R. L. Pictet and P. W. Morris 1973 Toward molecular mechanisms of developmental processes. Annu. Rev. Biochem., 42: 601-616. Sellers, L., and D. Granner 1974 Regulation of tyrosine aminotransferase activity in two liver-derived permanent cell lines. J. Cell. Biol., 60: 337-345. Sibley, C. H., and G. M. Tomkins 1974 Mechanisms of steroid resistance. Cell, 2: 221-227. Smith, J. A., and L. Martin 1973 Do cells cycle? Proc. Natl. Acad. Sci. (U.S.A.), 70: 1263-1267. Steinkamp, J. A,, M. J. Fulwyler, S. R. Coulter, R. D. Hie-
bert, S. L. Homey and P. F. Mullaney 1973 A new multiparameter separator for microscopic particles and biological cells. Rev. Sci. Instrum., 44: 1301-1310. Stellwagen, R. H., and G. M. Tomkins 1971 Preferential inhibition by 5-bromodeoxyuridine of the synthesis of tyrosine aminotransferase in hepatoma cell cultures. J. Mol. Biol., 56: 167-182. Stetten, G., S. A. Latt and R. L. Davidson 1976 33258 Hoechst enhancement of the photosensitivity of bromodeoxyuridine-substituted cells. Somatic Cell Genetics, 2: 285-290. Stuart, J. T., and M. Ingram 1971 The effect of cortisol on viability and glucose uptake in rat thymocytes in uitro. Proc. Soc. Exp. Biol. Med., 136: 1146-1150. Talavera, A., and C. Basilico 1977 Temperature sensitive mutants of BHK cells affected in cell cycle progression. J. Cell. Physiol., 92: 425-436. Yamamoto, K. R., M. R. Stampfer and G. M. Tomkins 1974 Receptors from glucocorticoid-sensitivelymphoma cells and two classes of insensitive clones: Physical and DNA binding properties. Proc. Natl. Acad. Sci. (U.S.A.), 21: 3901-3905.
