Effects of Cyctic AMP on the Growth of ~ ~ f ~ e r e n t ~and at~ng Undifferentiated Friend Erythroleukemic Cells CHARLES S . RUBIN Departments of Molecular Pharmacology and Neuroscience, Albert Einstein College of Medicine, Bronx, New York 10461
ABSTRACT Elevated concentrations of cyclic AMP elicit only minor reductions in growth rate and saturation density in undifferentiated Friend erythroleukemic cells. During the course of dimethylsulfoxide (DMSO)-induced differentiation, Friend cells convert from a cyclic AMP-tolerant state to a phenotype characterized by a high degree of sensitivity to cyclic AMP-mediated growth arrest. Conversion to cyclic AMP sensitivity is detectable after 30 hours growth in medium containing 2% DMSO, and either 0.5 mM 8-Br-cyclicAMP or 5 nM cholera toxin. Cultures of differentiating Friend cells achieved a stationary phase density that was approx~mate~y %fold higher than the cell density observed in parallel, differentiating cultures treated with 0.5 mM 8-Br-cyclic AMP. Temporally, the appearance of cyclic AMP-sensitivity corresponds to the early expression of in vitro erythroid differentiation (Ross et al., '741, but growth arrest does not al. ter the subsequent accumulation of hemoglobin in non-dividing DMSO-induced cells. Since growth arrest is preceded by a round of cell division, these observations are consistent with the concept that DMSO must be present during DNA replication for the subsequent expression of hemoglobin synthesis (McClintock and Papaconstantinou, '74; Levy et al., '75; Harrison, '76). Murine e~throleukemiccells infected with plasma membrane antigens (Ikawa et al,, '731 Friend virus maintain the capability of un- and a severely diminished capacity for condergoing a defined sequence of morphological tinuous growth and division (Friend et al., '71) (Sato e t al., 'TI.), biochemical (Friend et al., have also been observed in hemog~ob~n-con'71; Sassa, "761 and antigenic (Ikawa et al., taining, differentiated cells. These coordi'73) alterations in vitro that are characteris- nated biochemical phenomena that constitute tic of erythroid differentiation. Highly polar in vitro differentiation in Friend erythroleucompounds such as dimethylsulfoxide (Friend kemic cells parallel the events associated with et al., '711, 1-methyl-2-piperidone(Tanaka et the maturation of proerythroblasts into noral., '75) and butyric acid (Leder and Leder, '75) moblasts during the course of normal erythroa s well as hypoxanthine and purine analogs cyte development (Cantor et al., '72; Maniatis (Gusefla and Nousman, '76) activate a de- et al., '731, velopmental program that results in the synBiochemical investigations on differentiathesis of globin messenger RNA (mRNA1 ting erythroleukemic cells have focused pri(Ross et al., '72; Ross et al., '74), the induction marily on the regulation of globin gene expresof heme biosynthetic enzymes (Sassa, '761, in- sion (Friend et al., '71; Ross et al., '74). The creased iron uptake (Friend et al., '711, heme kinetics of globin mRNA accumulation and biosynthesis (Friend et al., '71; Sassa, '761, the globin chain synthesis have been described translation of globin mRNA (Boyer et al,, '72; and globin gene expression has been quantiOstertag et al., '72) and the accumulation of tated by hybridization of mRNA to globin hemoglobin to a level of 25% of total cell pro- complementary DNA (Ross et al., "72; Ross et tein (Ostertag, '72) over a period of four to six al., '74). days. The appearance of mature erythrocyte Received Apr. 14, '77. Accepted Aug. 3, '77. J. CELL.PWYSIOL. (1978)94: 57-68.
57
58
CHARLES S. RUBIN
Certain aspects of the enhanced rate of heme biosynthesis and the sequential induction of heme biosynthetic enzymes have been afforded some attention (Friend et al., '71; Sassa, '76). In brief reports, Reem and Friend ('75, '76) have noted diminished activities in nucleotide synthetic enzymes in differentiating erythroleukemic cells that parallel similar decrements in developing erythrocytes. Other potential aspects of cellular regulation in differentiating erythroleukemic cells have not been explored. Numerous studies indicate that adenosine 3': 5'-monophosphate (cyclic AMP) is a n important regulator of cell physiology and growth in normal and transformed fibroblasts and lymphoid cells (Pastan et al., '75). Regulation by cyclic AMP may be positive or, more commonly, negative (growth inhibitory) and its action is often correlated with a specific segment of the cell cycle (Pastan et al., '75; Pastan and Johnson, '74). Since Friend cells possess an adenylate cyclase that is stimulated by P-adrenergic agonists, prostaglandin El and cholera toxin as well as cyclic AMP-dependent protein kinase activity (table 11, investigations were undertaken to determine the effects of increasing cellular cyclic AMP levels on the rates of growth in undifferentiated and differentiating erythroleukemic cells. Studies presented in this communication show: (1) elevated intracellular concentrations of cyclic AMP elicit minor reductions in growth rate and saturation density in undifferentiated erythroleukemic cells, (2) during the course of DMSO-induced differentiation, Friend cells convert from a cyclic AMPtolerant state to a phenotype characterized by sensitivity to cyclic AMP-mediated growth arrest, (3) conversion to cyclic AMP-sensitivity is detectable after 24 to 30 hours of DMSO treatment and temporally corresponds to the early expression of the erythroid differentiation program (Ross et al., '72; Ross et al., '74; Sassa, '76) and (4) cyclic AMP-mediated growth arrest does not affect the accumulation of hemoglobin in the non-dividing cells. MATERIALS AND METHODS
GMP), uridine, cytidine, isoproterenol, epinephrine, norepinephrine, propranolol, prostaglandin El, creatine phosphate and creatine phosphokinase were purchased from Sigma Chemical Company. 8-Br-adenosine 3':5'monophosphate (8-Br-cyclic AMP) was obtained from Plenum (Hackensack, New Jersey). Cholera toxin and cyclic [Eb3H1AMP (22 Cifrnmole) w e r e p u r c h a s e d f r o m SchwartzIMann. 1-Methyl-3-isobutylxanthine (MIX) was obtained from Aldrich Chemical Company. Dimethylsulfoxide (DMSO) was procured from Baker Chemical Company. ATP (12 Ci/mmole) was obtained from ICN; [ Y - ~ ~ ATP P I (16 Ci/mmole) was from Amersham-Searle.
Cells Friend murine erythroleukemic cells (clone 745) were obtained from Doctor A r t Skoultchi, Department of Cell Biology, Albert Einstein College of Medicine. Friend cells were grown in Dulbecco's Modified Eagle's Medium (Gibco) containing 8.9%fetal calf serum (Gibco)in an atmosphere of 10% co2/90%air. Cultures were split and seeded a t 1.5-2.0 x 105/ml every 24 hours to maintain continuous log phase cells. Murine lymphosarcoma cells (S49) were obtained from the Salk Institute and were maintained under the conditions described above. All cell counts were performed manually using a hemacytometer. In the experiments shown in figures 2, 3, 4 and 6, cell viability was confirmed by plating efficiency determinations. Friend cell differentiationand hemoglobin staining Erythroid differentiation was induced by growing Friend cells in standard medium (see above) supplemented with 2% (v/v) DMSO (Friend et al., '71). Hemoglobin-containing cells were identified by treating 0.2 ml cells with 0.1 ml 10% acetic acid containing 0.3%benzidine base and 1.5% hydrogen peroxide (Orkin et al., '75). Cells containing hemoglobin immediately stained bright blue. These are designated B' (benzidine positive).
Materials Assays Adenosine 3':5'-monophosphate (cyclic Hemoglobin concentrations were estimated AMP), adenosine 5'-monophosphate (5'-AMP), adenosine 5'-triphosphate (ATP), guanosine spectrophotometrically a t 412 nm according 3':5'-monophosphate (cyclic GMP), 8-Br to the procedure described by McClintock and guanosine 3':5'-monophosphate (8-Br-cyclic Papaconstantinou ('74).
EFFECTS OF CYCLIC AMP ON FRIEND CELLS
Cyclic AMP concentrations were determined by the competitive binding method of Gilman ('70). Two minor modifications were made: 50 mM potassium phosphate buffer, pH 7.0 was substituted for sodium acetate and human erythrocyte plasma membranes (50 pg) were used as the source of the cyclic AMPbinding protein (Rubin et al., '72). The assay for cyclic nucleotide phosphodiesterase activity will be described in detail in another report (C. S. Rubin and R. Rangel, manuscript in preparation). In brief, the reaction mixture contained 40 mM Tris, pH 8.1,lO mM MgClz, 2 mM dithiothreitol, 1 p M cyclic [€L3H1 AMP (12,000 CPM/pmole), 50 fig bovine serum albumin and 2-10 p g enzyme preparation in a final volume of 50 p1. Assays were carried out a t 37" for ten minutes and were terminated by the addition of 5 ul of stopping solution, containing 0.2 MEDTA,pH 7.0, 12.5 mM adenosine, 12.5 mM 5'-AMP and 12.5 mM cyclic AMP. The percent hydrolysis of cyclic AMP to 5'-AMP and adenosine was determined by resolving the substrate and products by thin layer chromatography on PEI-cellulose plates (E. Merck), using 50 mM KCl as the solvent. Spots corresponding to cyclic AMP, 5'-AMP and adenosine were cut out, immersed in 8 ml Toluene/Omnifluor (NEN, 4 g,/L), and the radioactivity observed in adenosine and 5'-AMP is the total amount of cyclic AMP hydrolyzed. Adenylate cyclase was assayed under conditions described by Salomon et al. ('74). Assays were performed in 50 p1 of a mixture containing 25 mM Tris-HC1, pH 7.4, 5 mM MgC12, 1 mM cyclic AMP, 0.5 mM [a-32PlATP (150 cpm/pmole), 20 mM creatine phosphate, 2.5 p g creatine phosphokinase (0.4 units) and 40 p g membrane protein. After ten minutes a t 37", the reaction was terminated by the addition of 1 ml of a solution containing 1 pmole cyclic l3 HI AMP (30,000cpm), 1mM ATP and 1%sodium dodecyl sulfate. Cyclic l3?-P1AMP was isolated by chromatography on Dowex AG50 and alumina. The final column eluates were mixed with 3 volumes of Triton X-lOO/ toluene (3:7, v/v), containing 4 g Ominfluor (NEN) per liter, and the radioactivity was determined by scintillation spectrometry. The recovery of cyclic AMP was determined from the yield of cyclic L3H1AMP and the data were corrected accordingly. Cyclic AMP-dependent protein kinase activity was determined as described by Rubin e t al. ('74).
59
Preparation of crude membranes and cytosol for enzyme assay Friend cells (3 x loa) were harvested by sedimentation a t 200 x g for four minutes and then washed twice by resuspension in 15 ml of 0.15 M NaC1-5mM Tris, pH 7.4 and sedimentation a t 200 X g for four minutes. The cells were then allowed to swell in 15 volumes of 5 mM NaCl-5 mM Tris, pH 7.4, a t 0" for three minutes. All subsequent operations were at 0"-4".Subsequent to swelling, sufficient 0.25 M MgClz was added to achieve a final MgC1, concentration of 0.2 mM and the cells were then transferred to a Dounce homogenizer. Cell disruption was accomplished by 20 strokes of the "tight-fitting" pestle. Nuclei were pelleted from the homogenate by sedimentation a t 270 x g for three minutes, and the supernatant fluid was collected. The nuclear pellet was dispersed in 15 volumes of 5 mM NaC1-5 mM Tris, pH 7.4, and sedimented as described above. The resulting supernatant fluid was collected and combined with the previous supernatant and resolved into cytosol and crude membrane fractions by sedimentation a t 40,000 X g for ten minutes. The membranes were washed with 5 ml of 5 mM NaCl-5 mM Tris, pH 7.4, and resuspended in a volume equal to 50%of the volume of the cytosol. RESULTS
Adenylate cyclase and cyclic AMP-dependent protein kinase activities in differentiated and undfferentiated Friend cells
If cyclic AMP were to play a role in regulating Friend cell growth, then the cells would necessarily contain both a hormone-(or otherwise regulated adenylate cyclase and cyclic AMP-dependent protein kinase, the receptor for cyclic AMP in eukaryotic cells. Results of a typical set of adenylate cyclase and protein kinase assays are presented in table 1. Membranes from differentiated and undifferentiated cells possess approximately equal amounts of basal and fluoride-stimulated adenylate cyclase activities. Both cyclase preparations were stimulated 4 to 8-fold by the ,@-adrenergicagonists isoproterenol, epinephrine and norepinephrine. This cyclase activation was blocked by the ,@-antagonist, propranolol, while the a-blocker phentolamine had no effect. Maximal adenylate cyclase activities were observed in the presence of cholera toxin and prostaglandin El (PGE,).
60
CHARLES S. RUBIN TABLE 1 Adenylate cyclase activity pmoles cyclic AMP formediminlmg Addition
10 pM isoproterenol 10 pM epinephrine 10 pM norepinephrine 10pM isoproterenol + 10 pM propranolol 10pM isoproterenol 10 pM phentolamine 3 pM prostaglandin El 10 nM cholera toxin I 20 mM NaF
+
Undifferentiated cells
Differentiated cells
8.8 76.9 59.4 38.2 9.8
6.5 54.5 55.1 36.1 7.4
79.3
52.7
216.6 251.1 142.8
71.9 212.6 150.1
+ 2 p M cyclic AMP
0.78 1.44
0.59 1.61
Cytosol and membrane fractions were prepared from 3 X lo8 cells as described under MATERIALS AND METHODS. Adenylate cyclase assays were carried out as described under MATERIALS AND METHODS using the crude membrane fraction as the source of the enzyme. Seventyfive percent of the adenylate cyclase activity in cell homogenates was recovered in the membrane fractions from both differentiated and undifferentiated cells. Cytosol protein kinase activity was determined in the presence and absence of 2 pM cyclic AMP as indicated by Rubin et al. 1'74). 1 Cholera toxin-stimulated adenylate cyclase activity was measured in membranes isolated from cells that were pre-incubated for one hour at 37" in standard culture medium (MATERIALS AND METHODS) containing 10 nM cholera toxin. TABLE 2
Effect of cholera toxin on cyclic AMP levels in undifferentiated Friend erythroleukemic cells Cyclic AMP content Time of cholera toxin treatment ( m i d
0 45
60 90 960
P 2
-50
8
'ii \
N
>-
-40
pmotes/&g DNA pmoles/107 cells
0.06 4.2 4.0 4.7 2.1
2.5 217 222 235 126
Cells in the logarithmic phase of growth were treated with 5 nM cholera toxin and aliquots containing 2.5 X lo7 cells were removed from the cultures after the indicated time intervals. Cells were harvested by sedimentation and the cell pellets were extracted with 2 ml of 5% fw/v) trichloroacetic acid at 0'. Acid-soluble cyclic AMP was separated from precipitated macro-molecules by centrifugation at 2,000 x g for ten minutes. The supernatant was collected and extracted six times with 10 mi ethyl ether to remove trichloroacetic acid. When the pH of the sample was greater than 5, the cyclic AMP content was determined by the method of Gilman (Gilman, '70) with the modifications noted in MATERIALS AND METHODS. The DNA content of the pellet was determined by the method of Burton (Burton, '56).
$ n
-30 -20
Protein kinase activity nmoles =P transferred/min/mg
-
-60
-
'
' 0
20
40
60
80
100
120
40 J
0
10
t,
f
MIX CONCENTRATION ( p M )
Fig. 1 Inhibition of soluble and membrane-associated cyclic nucleotide phosphodiesterases by 1-methyl-3-isobutyl xanthine (MIX).Cytosol andmemhrane fractions were prepared as described under MATERIALS AND METHODS. Enzyme assays were performed as indicated under MATERIALS AND METHODS except that the indicated concentrations of MIX were included in the reaction mixtures. Specific activities for the soluble diesterase (0-0-0) aregivenon theleft ordinate; specific activitiesfor the membrane-associated enzyme (0--o--0) are on the right.
In general, adenylate cyclase activity in membranes from differentiated cells was comparable to the activity in membranes isolated from undifferentiated cells with the exception of the lower absolute value for PGE, -stimulation. Nevertheless, adenylate cyclase from the differentiated cells was still quite sensitive to 3 pM PGEl, being activated more than 10fold. Differentiated and undifferentiated Friend cells also possessed similar cyclic AMP-dependent protein kinase activities (table 1). Effects of cyclic AMP on undifferentiated Friend erythroleukemic cells Intracellular cyclic AMP levels were raised by the addition of exogenous 8-bromoadenosine 3'-5'-monophosphate (8-Br-cyclic AMP) or by eliciting the generation of cyclic AMP in situ by treating cultures with cholera toxin (Finkelstein, '73). 8-Br-cyclic AMP was chosen because this analog is 10 to 20-fold more resistant to cyclic nucleotide phosphodiesterase-mediatedhydrolysis than cyclic AMP (Simon et al., '73;
61
EFFECTS OF CYCLIC AMP ON FRIEND CELLS
4.0
3x)
-
-
2.0 -
Conc.of 8Br-CAMP 0-00
0-P
o----o 0.2 mm
).--.0.5mM 0-43 1.0mM A--AS49-WT,O.ZmM
Y
2
1.0 5: 0.8
--
0.1
0
Ib
2)o
o;
40
s'o
$0
'$4
TIME AFTER SEEDING (hours) Fig. 2 Effects of 8-Br-cyclic AMP on the growth of undifferentiated Friend cells. Cells were grown in Duibecco's modified Eagle's medium containing 8.9%fetal calf serum, 0.1 mM MIX and various concentrations of 8-Br-cyclicAMP 0.2 mM 8-Br-cyclicAMP (0- -0- -0),0.5 mM 8-Br-cyclicAMP (m-. .-4). 1.0 mM 8Br-cyclicAMP (0- -El ), control, no added 8-Br-cyclicAMP (0-0-0). S49 lymphoma cells (A-.-A)were grown under the same conditionsin the presence of 0.2 mM 8-Br-cyclicAMP. 8-Br-cyclicAMP was added at zero time. Aliquots of cells were removed and counted in hemacytometers at the indicated times.
Meyer and Miller, '74) and is equipotent with cyclic AMP in activating cyclic AMP-dependent protein kinases (Simon, et al., '73; Meyer and Miller, '741, the molecular receptors for cyclic AMP. Investigations by Leder and Leder ('75) have shown that butyric acid induces differentiation in some lines of erythroleukemic cells, thereby preclucing the use of N6,02-dibutyryl-adenosine3'-5'-monophosphate (a commonly studied cyclic AMP analog) in the present studies (DISCUSSION). 1-Methyl-3-isobutyl xanthine (MIX), an inhibitor of cyclic nucleotide phosphodiesterases (Beavo et al., '70) was used to potentiate the effects of 8-Br-cyclic AMP. Figure 1shows t h a t MIX was a potent inhibitor of both the soluble and membrane-associated phosphodiesterases of Friend cells, achieving nearly 90% inhibition at a concentration of 100 pM. This inhibition, coupled with the low rate of cyclic nucleotide phosphodiesterase-mediated hydrolysis of 8-Br-cyclic AMP (Simon et al., '76; Meyer and Miller, '741, insures the prolonged elevation of the intracellular cyclic nucleotide concentration during treatment with the 8.Br analog of cyclic AMP. In other experiments, the intracellular cyclic AMP concentration was raised by activating adenylate cyclase with cholera toxin (Finkelstein, '73; Kimberg et al., '71). Friend
cell adenylate cyclase was irreversibly activated 20 to 40-fold by exposing intact cells to 5-10 nM cholera toxin' for 45 minutes or longer (table 2). Adenylate cyclase activation persists for several days in cultured cells, provides a maximal supply of newly synthesized cyclic AMP which sustained a cyclic AMP level 25- to 100-fold greater than the cyclic AMP content of control cells (C. S. Rubin, manuscript in preparation). (See table 2.) When undifferentiated erythroleukemic cells in the logarithmic phase of growth were treated with several concentrations of 8-Brcyclic AMP only slight effects on growth rate were observed (fig. 2). The population doubling time increased from approximately 9.5 hours to a maximum of 1 2 hours in the presence of 1mM 8-Br-cyclicAMP, while the saturation density achieved after 84 hours dropped 20-30%. The relative insensitivity of Friend cells to cyclic AMP-mediated growth regulation is better appreciated by comparing these data to the effects of 8-Br-cyclicAMP on murine ,949 lymphoma cells (fig. 2; see Coffin0 e t al., '75; Bourne et al., '75). When exposed to 0.2 mM 8Br-cyclic AMP, S49 cells enter a state of arrested growth after 24 hours, and then proI Cholera toxin concentrationwas determinedby using a molecular weight of 84,000 (LoSpalluto and Finkelstein. '72).
62
CHARLES S. RUBIN
6.0 5.O
5.51-
4.0
3.O 2 .o
I.o
0.8 0.6 4 Control-no additions 0----a0.1 mM MIX .--.*I 2% DMSO 0--5 2% DMSO+O.lmM MIX
0.4 0.3
A--A 5nM Cholera toxin &---.-A5nM Cholera toxin+O.l rnMMlX
o ' 2 ~ 0.11 1
0
1
20
1
1
1
1
40
60
1
1
80
TIME AFTER SEEDING (hours) Fig. 3 Effects of MIX, cholera toxin and DMSO on the growth of Friend cells. Cells were grown and samped a s described in MATERIALS AND METHODS and figure 2. The indicated additions were made a t zero time. Cultures grown in t h e presence of 2%DMSO contained 90%differentiated cells 110 hours after seeding; the other four cultures contained undifferentiated cells.
3-01
2 .o
I
0.3
0 '
/-
TIME AFTER SEEDING (hours) Fig. 4 Effects of 8-Br-cyclic AMP and cholera toxin on the growth of differentiating erythroleukemic cells. Cells were grown and sampled a s described in MATERIALS AND METHODS and figure 2 except that 2% (v/v) DMSO was included in the medium. Cultures were treated with saline (control) (0-0-O), 1 mM 8-Br-cyclic AMP (O---O---O) and 10 nM cholera toxin (m-. .-.).
I
0
I
I
100
I
I
200
I
I
300
I
L
400
I
I
500
8Br-CAMP CONCENTRATION (pM) Fig. 5 Effect of varying concentrations of 8-Br-cyclic AMP on the cell density of differentiating Friend cells. Cells were seeded a t an initial density of 2.3 X W / m l in medium containing 2%DMSO, 0.1 mM MIX and the indicated concentrations of 8-Br-cyclicAMP. Cultures were grown for 132 hours to allow the cells to enter the stationary phase and express the differentiated phenotype a t the translational level (hemoglobin synthesis). Aliquots were then removed for t h e determination of cell density and the detection of hemoglobin synthesis by benzidine staining (MATERIALS AND METHODS). All cultures contained more than 86%B+ cells.
ceed through a period of metabolic decline that culminates in cell death and cytolysis after 72 to 96 hours (fig. 2; Coffin0 et al., '75). In analogous experiments, Friend cells were treated with 5 nM cholera toxin and the generation time was lengthened to 13 hours (fig. 3). This effect was enhanced by MIX after two population doublings. Lower concentrations of cholera toxin had small, but significant effects on growth. The ineffectiveness of cholera toxin in inhibiting growth cannot be ascribed to the absence of an appropriate ganglioside receptor or the failure to activate adenylate cyclase. Erythroleukemic cells increased their cyclic AMP content approximately 90-fold subsequent to the addition of 5 nM cholera toxin and sustained a cyclic AMP level 50-fold greater than untreated cells over a period that exceeds the generation time by three hours (table 2). The growth of Friend cells was also unpM) and prosaffected by isoproterenol (5-500 pg/ml). taglandin E, (0.01-1 * Differentiationwas monitoredby identifying hemoglobincontaining cells using the benzidine staining technique (Orkin et al., '75; MATERIALS AND METHODS).
63
EFFECTS OF CYCLIC AMP ON FRIEND CELLS TABLE 3
Appearance and quantitation of hemoglobin in differentiating Friend cells % B' cells
Additions
2%saline 2%DMSO 2%DMSO 0.5 mM 8 Br-cyclic AMP
+
Absorbance at 415 nm/lO' cells
74 hrs
84 hrs
96 hrs
120 hrs
0.01 0.31 0.24
ABSTRACT Elevated concentrations of cyclic AMP elicit only minor reductions in growth rate and saturation density in undifferentiated Friend erythroleukemic cells. During the course of dimethylsulfoxide (DMSO)-induced differentiation, Friend cells convert from a cyclic AMP-tolerant state to a phenotype characterized by a high degree of sensitivity to cyclic AMP-mediated growth arrest. Conversion to cyclic AMP sensitivity is detectable after 30 hours growth in medium containing 2% DMSO, and either 0.5 mM 8-Br-cyclicAMP or 5 nM cholera toxin. Cultures of differentiating Friend cells achieved a stationary phase density that was approx~mate~y %fold higher than the cell density observed in parallel, differentiating cultures treated with 0.5 mM 8-Br-cyclic AMP. Temporally, the appearance of cyclic AMP-sensitivity corresponds to the early expression of in vitro erythroid differentiation (Ross et al., '741, but growth arrest does not al. ter the subsequent accumulation of hemoglobin in non-dividing DMSO-induced cells. Since growth arrest is preceded by a round of cell division, these observations are consistent with the concept that DMSO must be present during DNA replication for the subsequent expression of hemoglobin synthesis (McClintock and Papaconstantinou, '74; Levy et al., '75; Harrison, '76). Murine e~throleukemiccells infected with plasma membrane antigens (Ikawa et al,, '731 Friend virus maintain the capability of un- and a severely diminished capacity for condergoing a defined sequence of morphological tinuous growth and division (Friend et al., '71) (Sato e t al., 'TI.), biochemical (Friend et al., have also been observed in hemog~ob~n-con'71; Sassa, "761 and antigenic (Ikawa et al., taining, differentiated cells. These coordi'73) alterations in vitro that are characteris- nated biochemical phenomena that constitute tic of erythroid differentiation. Highly polar in vitro differentiation in Friend erythroleucompounds such as dimethylsulfoxide (Friend kemic cells parallel the events associated with et al., '711, 1-methyl-2-piperidone(Tanaka et the maturation of proerythroblasts into noral., '75) and butyric acid (Leder and Leder, '75) moblasts during the course of normal erythroa s well as hypoxanthine and purine analogs cyte development (Cantor et al., '72; Maniatis (Gusefla and Nousman, '76) activate a de- et al., '731, velopmental program that results in the synBiochemical investigations on differentiathesis of globin messenger RNA (mRNA1 ting erythroleukemic cells have focused pri(Ross et al., '72; Ross et al., '74), the induction marily on the regulation of globin gene expresof heme biosynthetic enzymes (Sassa, '761, in- sion (Friend et al., '71; Ross et al., '74). The creased iron uptake (Friend et al., '711, heme kinetics of globin mRNA accumulation and biosynthesis (Friend et al., '71; Sassa, '761, the globin chain synthesis have been described translation of globin mRNA (Boyer et al,, '72; and globin gene expression has been quantiOstertag et al., '72) and the accumulation of tated by hybridization of mRNA to globin hemoglobin to a level of 25% of total cell pro- complementary DNA (Ross et al., "72; Ross et tein (Ostertag, '72) over a period of four to six al., '74). days. The appearance of mature erythrocyte Received Apr. 14, '77. Accepted Aug. 3, '77. J. CELL.PWYSIOL. (1978)94: 57-68.
57
58
CHARLES S. RUBIN
Certain aspects of the enhanced rate of heme biosynthesis and the sequential induction of heme biosynthetic enzymes have been afforded some attention (Friend et al., '71; Sassa, '76). In brief reports, Reem and Friend ('75, '76) have noted diminished activities in nucleotide synthetic enzymes in differentiating erythroleukemic cells that parallel similar decrements in developing erythrocytes. Other potential aspects of cellular regulation in differentiating erythroleukemic cells have not been explored. Numerous studies indicate that adenosine 3': 5'-monophosphate (cyclic AMP) is a n important regulator of cell physiology and growth in normal and transformed fibroblasts and lymphoid cells (Pastan et al., '75). Regulation by cyclic AMP may be positive or, more commonly, negative (growth inhibitory) and its action is often correlated with a specific segment of the cell cycle (Pastan et al., '75; Pastan and Johnson, '74). Since Friend cells possess an adenylate cyclase that is stimulated by P-adrenergic agonists, prostaglandin El and cholera toxin as well as cyclic AMP-dependent protein kinase activity (table 11, investigations were undertaken to determine the effects of increasing cellular cyclic AMP levels on the rates of growth in undifferentiated and differentiating erythroleukemic cells. Studies presented in this communication show: (1) elevated intracellular concentrations of cyclic AMP elicit minor reductions in growth rate and saturation density in undifferentiated erythroleukemic cells, (2) during the course of DMSO-induced differentiation, Friend cells convert from a cyclic AMPtolerant state to a phenotype characterized by sensitivity to cyclic AMP-mediated growth arrest, (3) conversion to cyclic AMP-sensitivity is detectable after 24 to 30 hours of DMSO treatment and temporally corresponds to the early expression of the erythroid differentiation program (Ross et al., '72; Ross et al., '74; Sassa, '76) and (4) cyclic AMP-mediated growth arrest does not affect the accumulation of hemoglobin in the non-dividing cells. MATERIALS AND METHODS
GMP), uridine, cytidine, isoproterenol, epinephrine, norepinephrine, propranolol, prostaglandin El, creatine phosphate and creatine phosphokinase were purchased from Sigma Chemical Company. 8-Br-adenosine 3':5'monophosphate (8-Br-cyclic AMP) was obtained from Plenum (Hackensack, New Jersey). Cholera toxin and cyclic [Eb3H1AMP (22 Cifrnmole) w e r e p u r c h a s e d f r o m SchwartzIMann. 1-Methyl-3-isobutylxanthine (MIX) was obtained from Aldrich Chemical Company. Dimethylsulfoxide (DMSO) was procured from Baker Chemical Company. ATP (12 Ci/mmole) was obtained from ICN; [ Y - ~ ~ ATP P I (16 Ci/mmole) was from Amersham-Searle.
Cells Friend murine erythroleukemic cells (clone 745) were obtained from Doctor A r t Skoultchi, Department of Cell Biology, Albert Einstein College of Medicine. Friend cells were grown in Dulbecco's Modified Eagle's Medium (Gibco) containing 8.9%fetal calf serum (Gibco)in an atmosphere of 10% co2/90%air. Cultures were split and seeded a t 1.5-2.0 x 105/ml every 24 hours to maintain continuous log phase cells. Murine lymphosarcoma cells (S49) were obtained from the Salk Institute and were maintained under the conditions described above. All cell counts were performed manually using a hemacytometer. In the experiments shown in figures 2, 3, 4 and 6, cell viability was confirmed by plating efficiency determinations. Friend cell differentiationand hemoglobin staining Erythroid differentiation was induced by growing Friend cells in standard medium (see above) supplemented with 2% (v/v) DMSO (Friend et al., '71). Hemoglobin-containing cells were identified by treating 0.2 ml cells with 0.1 ml 10% acetic acid containing 0.3%benzidine base and 1.5% hydrogen peroxide (Orkin et al., '75). Cells containing hemoglobin immediately stained bright blue. These are designated B' (benzidine positive).
Materials Assays Adenosine 3':5'-monophosphate (cyclic Hemoglobin concentrations were estimated AMP), adenosine 5'-monophosphate (5'-AMP), adenosine 5'-triphosphate (ATP), guanosine spectrophotometrically a t 412 nm according 3':5'-monophosphate (cyclic GMP), 8-Br to the procedure described by McClintock and guanosine 3':5'-monophosphate (8-Br-cyclic Papaconstantinou ('74).
EFFECTS OF CYCLIC AMP ON FRIEND CELLS
Cyclic AMP concentrations were determined by the competitive binding method of Gilman ('70). Two minor modifications were made: 50 mM potassium phosphate buffer, pH 7.0 was substituted for sodium acetate and human erythrocyte plasma membranes (50 pg) were used as the source of the cyclic AMPbinding protein (Rubin et al., '72). The assay for cyclic nucleotide phosphodiesterase activity will be described in detail in another report (C. S. Rubin and R. Rangel, manuscript in preparation). In brief, the reaction mixture contained 40 mM Tris, pH 8.1,lO mM MgClz, 2 mM dithiothreitol, 1 p M cyclic [€L3H1 AMP (12,000 CPM/pmole), 50 fig bovine serum albumin and 2-10 p g enzyme preparation in a final volume of 50 p1. Assays were carried out a t 37" for ten minutes and were terminated by the addition of 5 ul of stopping solution, containing 0.2 MEDTA,pH 7.0, 12.5 mM adenosine, 12.5 mM 5'-AMP and 12.5 mM cyclic AMP. The percent hydrolysis of cyclic AMP to 5'-AMP and adenosine was determined by resolving the substrate and products by thin layer chromatography on PEI-cellulose plates (E. Merck), using 50 mM KCl as the solvent. Spots corresponding to cyclic AMP, 5'-AMP and adenosine were cut out, immersed in 8 ml Toluene/Omnifluor (NEN, 4 g,/L), and the radioactivity observed in adenosine and 5'-AMP is the total amount of cyclic AMP hydrolyzed. Adenylate cyclase was assayed under conditions described by Salomon et al. ('74). Assays were performed in 50 p1 of a mixture containing 25 mM Tris-HC1, pH 7.4, 5 mM MgC12, 1 mM cyclic AMP, 0.5 mM [a-32PlATP (150 cpm/pmole), 20 mM creatine phosphate, 2.5 p g creatine phosphokinase (0.4 units) and 40 p g membrane protein. After ten minutes a t 37", the reaction was terminated by the addition of 1 ml of a solution containing 1 pmole cyclic l3 HI AMP (30,000cpm), 1mM ATP and 1%sodium dodecyl sulfate. Cyclic l3?-P1AMP was isolated by chromatography on Dowex AG50 and alumina. The final column eluates were mixed with 3 volumes of Triton X-lOO/ toluene (3:7, v/v), containing 4 g Ominfluor (NEN) per liter, and the radioactivity was determined by scintillation spectrometry. The recovery of cyclic AMP was determined from the yield of cyclic L3H1AMP and the data were corrected accordingly. Cyclic AMP-dependent protein kinase activity was determined as described by Rubin e t al. ('74).
59
Preparation of crude membranes and cytosol for enzyme assay Friend cells (3 x loa) were harvested by sedimentation a t 200 x g for four minutes and then washed twice by resuspension in 15 ml of 0.15 M NaC1-5mM Tris, pH 7.4 and sedimentation a t 200 X g for four minutes. The cells were then allowed to swell in 15 volumes of 5 mM NaCl-5 mM Tris, pH 7.4, a t 0" for three minutes. All subsequent operations were at 0"-4".Subsequent to swelling, sufficient 0.25 M MgClz was added to achieve a final MgC1, concentration of 0.2 mM and the cells were then transferred to a Dounce homogenizer. Cell disruption was accomplished by 20 strokes of the "tight-fitting" pestle. Nuclei were pelleted from the homogenate by sedimentation a t 270 x g for three minutes, and the supernatant fluid was collected. The nuclear pellet was dispersed in 15 volumes of 5 mM NaC1-5 mM Tris, pH 7.4, and sedimented as described above. The resulting supernatant fluid was collected and combined with the previous supernatant and resolved into cytosol and crude membrane fractions by sedimentation a t 40,000 X g for ten minutes. The membranes were washed with 5 ml of 5 mM NaCl-5 mM Tris, pH 7.4, and resuspended in a volume equal to 50%of the volume of the cytosol. RESULTS
Adenylate cyclase and cyclic AMP-dependent protein kinase activities in differentiated and undfferentiated Friend cells
If cyclic AMP were to play a role in regulating Friend cell growth, then the cells would necessarily contain both a hormone-(or otherwise regulated adenylate cyclase and cyclic AMP-dependent protein kinase, the receptor for cyclic AMP in eukaryotic cells. Results of a typical set of adenylate cyclase and protein kinase assays are presented in table 1. Membranes from differentiated and undifferentiated cells possess approximately equal amounts of basal and fluoride-stimulated adenylate cyclase activities. Both cyclase preparations were stimulated 4 to 8-fold by the ,@-adrenergicagonists isoproterenol, epinephrine and norepinephrine. This cyclase activation was blocked by the ,@-antagonist, propranolol, while the a-blocker phentolamine had no effect. Maximal adenylate cyclase activities were observed in the presence of cholera toxin and prostaglandin El (PGE,).
60
CHARLES S. RUBIN TABLE 1 Adenylate cyclase activity pmoles cyclic AMP formediminlmg Addition
10 pM isoproterenol 10 pM epinephrine 10 pM norepinephrine 10pM isoproterenol + 10 pM propranolol 10pM isoproterenol 10 pM phentolamine 3 pM prostaglandin El 10 nM cholera toxin I 20 mM NaF
+
Undifferentiated cells
Differentiated cells
8.8 76.9 59.4 38.2 9.8
6.5 54.5 55.1 36.1 7.4
79.3
52.7
216.6 251.1 142.8
71.9 212.6 150.1
+ 2 p M cyclic AMP
0.78 1.44
0.59 1.61
Cytosol and membrane fractions were prepared from 3 X lo8 cells as described under MATERIALS AND METHODS. Adenylate cyclase assays were carried out as described under MATERIALS AND METHODS using the crude membrane fraction as the source of the enzyme. Seventyfive percent of the adenylate cyclase activity in cell homogenates was recovered in the membrane fractions from both differentiated and undifferentiated cells. Cytosol protein kinase activity was determined in the presence and absence of 2 pM cyclic AMP as indicated by Rubin et al. 1'74). 1 Cholera toxin-stimulated adenylate cyclase activity was measured in membranes isolated from cells that were pre-incubated for one hour at 37" in standard culture medium (MATERIALS AND METHODS) containing 10 nM cholera toxin. TABLE 2
Effect of cholera toxin on cyclic AMP levels in undifferentiated Friend erythroleukemic cells Cyclic AMP content Time of cholera toxin treatment ( m i d
0 45
60 90 960
P 2
-50
8
'ii \
N
>-
-40
pmotes/&g DNA pmoles/107 cells
0.06 4.2 4.0 4.7 2.1
2.5 217 222 235 126
Cells in the logarithmic phase of growth were treated with 5 nM cholera toxin and aliquots containing 2.5 X lo7 cells were removed from the cultures after the indicated time intervals. Cells were harvested by sedimentation and the cell pellets were extracted with 2 ml of 5% fw/v) trichloroacetic acid at 0'. Acid-soluble cyclic AMP was separated from precipitated macro-molecules by centrifugation at 2,000 x g for ten minutes. The supernatant was collected and extracted six times with 10 mi ethyl ether to remove trichloroacetic acid. When the pH of the sample was greater than 5, the cyclic AMP content was determined by the method of Gilman (Gilman, '70) with the modifications noted in MATERIALS AND METHODS. The DNA content of the pellet was determined by the method of Burton (Burton, '56).
$ n
-30 -20
Protein kinase activity nmoles =P transferred/min/mg
-
-60
-
'
' 0
20
40
60
80
100
120
40 J
0
10
t,
f
MIX CONCENTRATION ( p M )
Fig. 1 Inhibition of soluble and membrane-associated cyclic nucleotide phosphodiesterases by 1-methyl-3-isobutyl xanthine (MIX).Cytosol andmemhrane fractions were prepared as described under MATERIALS AND METHODS. Enzyme assays were performed as indicated under MATERIALS AND METHODS except that the indicated concentrations of MIX were included in the reaction mixtures. Specific activities for the soluble diesterase (0-0-0) aregivenon theleft ordinate; specific activitiesfor the membrane-associated enzyme (0--o--0) are on the right.
In general, adenylate cyclase activity in membranes from differentiated cells was comparable to the activity in membranes isolated from undifferentiated cells with the exception of the lower absolute value for PGE, -stimulation. Nevertheless, adenylate cyclase from the differentiated cells was still quite sensitive to 3 pM PGEl, being activated more than 10fold. Differentiated and undifferentiated Friend cells also possessed similar cyclic AMP-dependent protein kinase activities (table 1). Effects of cyclic AMP on undifferentiated Friend erythroleukemic cells Intracellular cyclic AMP levels were raised by the addition of exogenous 8-bromoadenosine 3'-5'-monophosphate (8-Br-cyclic AMP) or by eliciting the generation of cyclic AMP in situ by treating cultures with cholera toxin (Finkelstein, '73). 8-Br-cyclic AMP was chosen because this analog is 10 to 20-fold more resistant to cyclic nucleotide phosphodiesterase-mediatedhydrolysis than cyclic AMP (Simon et al., '73;
61
EFFECTS OF CYCLIC AMP ON FRIEND CELLS
4.0
3x)
-
-
2.0 -
Conc.of 8Br-CAMP 0-00
0-P
o----o 0.2 mm
).--.0.5mM 0-43 1.0mM A--AS49-WT,O.ZmM
Y
2
1.0 5: 0.8
--
0.1
0
Ib
2)o
o;
40
s'o
$0
'$4
TIME AFTER SEEDING (hours) Fig. 2 Effects of 8-Br-cyclic AMP on the growth of undifferentiated Friend cells. Cells were grown in Duibecco's modified Eagle's medium containing 8.9%fetal calf serum, 0.1 mM MIX and various concentrations of 8-Br-cyclicAMP 0.2 mM 8-Br-cyclicAMP (0- -0- -0),0.5 mM 8-Br-cyclicAMP (m-. .-4). 1.0 mM 8Br-cyclicAMP (0- -El ), control, no added 8-Br-cyclicAMP (0-0-0). S49 lymphoma cells (A-.-A)were grown under the same conditionsin the presence of 0.2 mM 8-Br-cyclicAMP. 8-Br-cyclicAMP was added at zero time. Aliquots of cells were removed and counted in hemacytometers at the indicated times.
Meyer and Miller, '74) and is equipotent with cyclic AMP in activating cyclic AMP-dependent protein kinases (Simon, et al., '73; Meyer and Miller, '741, the molecular receptors for cyclic AMP. Investigations by Leder and Leder ('75) have shown that butyric acid induces differentiation in some lines of erythroleukemic cells, thereby preclucing the use of N6,02-dibutyryl-adenosine3'-5'-monophosphate (a commonly studied cyclic AMP analog) in the present studies (DISCUSSION). 1-Methyl-3-isobutyl xanthine (MIX), an inhibitor of cyclic nucleotide phosphodiesterases (Beavo et al., '70) was used to potentiate the effects of 8-Br-cyclic AMP. Figure 1shows t h a t MIX was a potent inhibitor of both the soluble and membrane-associated phosphodiesterases of Friend cells, achieving nearly 90% inhibition at a concentration of 100 pM. This inhibition, coupled with the low rate of cyclic nucleotide phosphodiesterase-mediated hydrolysis of 8-Br-cyclic AMP (Simon et al., '76; Meyer and Miller, '741, insures the prolonged elevation of the intracellular cyclic nucleotide concentration during treatment with the 8.Br analog of cyclic AMP. In other experiments, the intracellular cyclic AMP concentration was raised by activating adenylate cyclase with cholera toxin (Finkelstein, '73; Kimberg et al., '71). Friend
cell adenylate cyclase was irreversibly activated 20 to 40-fold by exposing intact cells to 5-10 nM cholera toxin' for 45 minutes or longer (table 2). Adenylate cyclase activation persists for several days in cultured cells, provides a maximal supply of newly synthesized cyclic AMP which sustained a cyclic AMP level 25- to 100-fold greater than the cyclic AMP content of control cells (C. S. Rubin, manuscript in preparation). (See table 2.) When undifferentiated erythroleukemic cells in the logarithmic phase of growth were treated with several concentrations of 8-Brcyclic AMP only slight effects on growth rate were observed (fig. 2). The population doubling time increased from approximately 9.5 hours to a maximum of 1 2 hours in the presence of 1mM 8-Br-cyclicAMP, while the saturation density achieved after 84 hours dropped 20-30%. The relative insensitivity of Friend cells to cyclic AMP-mediated growth regulation is better appreciated by comparing these data to the effects of 8-Br-cyclicAMP on murine ,949 lymphoma cells (fig. 2; see Coffin0 e t al., '75; Bourne et al., '75). When exposed to 0.2 mM 8Br-cyclic AMP, S49 cells enter a state of arrested growth after 24 hours, and then proI Cholera toxin concentrationwas determinedby using a molecular weight of 84,000 (LoSpalluto and Finkelstein. '72).
62
CHARLES S. RUBIN
6.0 5.O
5.51-
4.0
3.O 2 .o
I.o
0.8 0.6 4 Control-no additions 0----a0.1 mM MIX .--.*I 2% DMSO 0--5 2% DMSO+O.lmM MIX
0.4 0.3
A--A 5nM Cholera toxin &---.-A5nM Cholera toxin+O.l rnMMlX
o ' 2 ~ 0.11 1
0
1
20
1
1
1
1
40
60
1
1
80
TIME AFTER SEEDING (hours) Fig. 3 Effects of MIX, cholera toxin and DMSO on the growth of Friend cells. Cells were grown and samped a s described in MATERIALS AND METHODS and figure 2. The indicated additions were made a t zero time. Cultures grown in t h e presence of 2%DMSO contained 90%differentiated cells 110 hours after seeding; the other four cultures contained undifferentiated cells.
3-01
2 .o
I
0.3
0 '
/-
TIME AFTER SEEDING (hours) Fig. 4 Effects of 8-Br-cyclic AMP and cholera toxin on the growth of differentiating erythroleukemic cells. Cells were grown and sampled a s described in MATERIALS AND METHODS and figure 2 except that 2% (v/v) DMSO was included in the medium. Cultures were treated with saline (control) (0-0-O), 1 mM 8-Br-cyclic AMP (O---O---O) and 10 nM cholera toxin (m-. .-.).
I
0
I
I
100
I
I
200
I
I
300
I
L
400
I
I
500
8Br-CAMP CONCENTRATION (pM) Fig. 5 Effect of varying concentrations of 8-Br-cyclic AMP on the cell density of differentiating Friend cells. Cells were seeded a t an initial density of 2.3 X W / m l in medium containing 2%DMSO, 0.1 mM MIX and the indicated concentrations of 8-Br-cyclicAMP. Cultures were grown for 132 hours to allow the cells to enter the stationary phase and express the differentiated phenotype a t the translational level (hemoglobin synthesis). Aliquots were then removed for t h e determination of cell density and the detection of hemoglobin synthesis by benzidine staining (MATERIALS AND METHODS). All cultures contained more than 86%B+ cells.
ceed through a period of metabolic decline that culminates in cell death and cytolysis after 72 to 96 hours (fig. 2; Coffin0 et al., '75). In analogous experiments, Friend cells were treated with 5 nM cholera toxin and the generation time was lengthened to 13 hours (fig. 3). This effect was enhanced by MIX after two population doublings. Lower concentrations of cholera toxin had small, but significant effects on growth. The ineffectiveness of cholera toxin in inhibiting growth cannot be ascribed to the absence of an appropriate ganglioside receptor or the failure to activate adenylate cyclase. Erythroleukemic cells increased their cyclic AMP content approximately 90-fold subsequent to the addition of 5 nM cholera toxin and sustained a cyclic AMP level 50-fold greater than untreated cells over a period that exceeds the generation time by three hours (table 2). The growth of Friend cells was also unpM) and prosaffected by isoproterenol (5-500 pg/ml). taglandin E, (0.01-1 * Differentiationwas monitoredby identifying hemoglobincontaining cells using the benzidine staining technique (Orkin et al., '75; MATERIALS AND METHODS).
63
EFFECTS OF CYCLIC AMP ON FRIEND CELLS TABLE 3
Appearance and quantitation of hemoglobin in differentiating Friend cells % B' cells
Additions
2%saline 2%DMSO 2%DMSO 0.5 mM 8 Br-cyclic AMP
+
Absorbance at 415 nm/lO' cells
74 hrs
84 hrs
96 hrs
120 hrs
0.01 0.31 0.24
