0 1991 Wiley-Liss, Inc.
Cytometry 12:119-126 (1991)
Method for Kinetic Analysis of Drug-Induced Cell Cycle Perturbations Paolo Ubezio, Stefania Filippeschi, and Laura Spinelli Istituto di Ricerche Farmacologiche “Mario Negri,” 20157 Milano, Italy Received for publication May 21, 1990; accepted October 9, 1990
A method is described for quantitative study of the flux of cells through the cell cycle phases in in vitro systems perturbed by chemicals, such as chemotherapeutic agents. The method utilizes cell count and the flow cytometric technique of bromodeoxyuridine (BrdUrd)labeling, according to an optimized strategy. Cells are exposed to BrdUrd during the last minutes of drug treatment and fixed for analysis at 0,1/3T,, 2/3T,, T,, and T, + TG1 recovery times, where T,, TG1, T, are the mean durations of phases S and G1 and of the whole cycle of control cells. As an example of application of the proposed procedure, a kinetic study of the effect of l-(2-chloroethyl)-l-nitrosourea (CNU) on the L1210 cell cycle is described. Simple
Since the introduction in flow cytometry of the 5bromodeoxyuridine (BrdUrd) technique, the method has been recognized as a powerful tool for studying cell kinetics (3,8,10,14,15). However, the exact relationship between the flow cytometric (FCM) data obtained with this method and the true kinetic picture of a proliferating cell population is not simple. Only in a few studies have complex cell cycle models been used to fit FCM data (5,9,19). In most studies, only FCM data are presented, with no further attempt to obtain quantitative information through higher level kinetic data analysis. We devised a n optimized strategy that enabled us to obtain quantitative data on the flux of cells through cell cycle phases, using a minimum of FCM measurements with the propidium iodide (PI)/anti-BrdUrd method and simple data analysis. This analysis is approached in two steps. The first requires oriented analysis of biparametric DNA/antiBrdUrd distributions; this can be done as a “window” selection by all commercial computers fitted to flow cytometers. The second step requires only a pocket calculator to make the appropriate counts on first step FCM data. Cell numbers must be obtained by independent parallel measurements. The use of the method is illustrated by a n
data analysis, requiring only a pocket calculator, showed that cells in phases G1 and G2M at the end of a 1h treatment with 1pg/ml CNU were fully able to leave these phases but were destined to remain blocked in the following G2M phase (G1 for a minority of them). We also found that cells initially in S phase were slightly delayed in completing their S phase and that 50% of them remained temporarily blocked in the subsequent G2M phase, irrespective of their position in the S phase. Key terms: Antibromodeoxyuridine, flow cytometry, nitrosourea compounds, cell kinetics, dual-parameter analysis
experiment in which the cell cycle kinetic perturbation induced in vitro by a n anticancer drug is measured quantitatively, and the different effects in different phases are distinguished simultaneously.
MATERIALS AND METHODS Cell Culture L1210 murine leukemia cells were grown in RPMI 1640 medium (Gibco Europe, Glasgow, Scotland) supplemented with 20% heat-inactivated (56°C for 30 min) fetal calf serum (FCS; Flow Laboratories, Irvine, Scotland) as a suspension under standard culture conditions, at a density between 0.1 x lo6 and 0.5 x l o 6 celldm1 subcultured every 3 or 4 days. Drug Treatment l-(2-chloroethyl)-l-nitrosourea(CNU) was kindly provided by the Drug Synthesis and Chemistry Branch, Drug Therapeutics Program, National Cancer Institute (Bethesda, MD). Drug solutions in 95% ethanol were prepared immediately before use. The final ethanol concentration in medium after treatment never exceeded 0.01%.
UBEZIO ET AL.
120
500
0
6
12
Cell growth Before the kinetic study of drug-induced cell cycle perturbation, the kinetics of exponentially growing cells had to be studied to obtain information necessary for setting the experimental design. The in vitro growth curve was studied, and the best conditions of medium, serum, and seeding concentration were established to obtain reproducible balanced exponential growth over periods exceeding the duration of the drug experiment (about two doubling in our case). CNUlW/ml I
18
Time (h)
FIG.1. Growth of L1210 cells during experimental time. Control cells and cells treated with 1 pgiml CNU for 1 h before time 0. The growth curve data have been normalized to give the initial value of 1,000 cells/ml at time 0.
Label With BrdUrd Cells were seeded at 0.02 x lo6 cellsiml with 20% FCS and incubated for 48 h, then treated with 1 pg/ml of CNU for 1h at 37°C. The growth inhibition produced by the treatment is shown in Figure 1. In the last 15 min of treatment, cells were also incubated with 5 p M BrdUrd (Sigma). At the end of the incubation cells were washed in phosphate-buffered saline (PBS), pH 7.4, centrifuged, and resuspended in fresh medium. At the specific recovery times, cells were counted using a Coulter counter (Coulter Electronics Ltd., Harpenden, Herts, U.K.) equipped with a size analyzer (Coulter Channalyzer C-lOOO), fixed in 70% ethanol, and stored a t 4°C. PI-anti-BrdUrdStaining Technique The cells were stained by PI and anti-BrdUrd direct immunofluorescent method, slightly modifying the method of Dolbeare et al. (2). The ethanol-fixed cell suspensions were centrifuged and incubated with 3 N HC1 for 20 min, to obtain partially denatured DNA (7,ll). After washing with 0.1 M Na,B,O, to neutralize the acid, the cell pellets were resuspended with 50 pl of Tween 20 (Merck) 0.5% in PBS. After that, 50 p.1 of bovine serum albumin 0.5% in PBS and 20 pl of fluoresceinated anti-BrdUrd (4,6,17) (Becton Dickinson, Mountain View, CA) were added to each tube and incubated for 30 min a t room temperature. At the end of the antibody exposure, the cells were centrifuged, resuspended in 1 pg/ml of PI in PBS, and analyzed after a t least 1 h. Flow Cytometric Measurements A Facstar (Becton Dickinson) was used. Argon ion laser excitation was 100 mW at the 488 nm line. Green fluorescein fluorescence was detected in the 515-555 nm wavelength band and red PI fluorescence above 620 nm.
Basic Considerations The duration of cell cycle phases (TG1,T,, TGBM) during exponential growth was estimated from the percentage of cells in the cell cycle phases (%G1, %S, %G2M, which are constant in balanced growth), as detected by standard DNA-FCM methods and from the doubling time (TD),obtained by cell counts, using the following formulae (13): TCZM= Zog(2)-' x T D x log(1
(T,
+ %G2M/100)
(1)
+
TGBM) = Zog(2)r' x T D x Zog(1 + [%S + %G2Ml/100) T,,
=
TD - ( T , + T G 2 d .
(2)
(3)
Using these formulae, which are derived from the hypothesis of absence of quiescent or lost cells, and neglecting intercell variability, with our experimental model we obtained the following values: T, = 12 h, T,, = 2.5 h , T, = 7.5 h, and TGZM = 2 h. Previous work from our laboratory (16) showed that equations 1-3 hold with good approximation even in the presence of high intercell variability (CV of 50%), reinterpreting TG1, T,, and TGZM as mean values of the respective phase durations. In the presence of quiescent cells, we adopted the procedure shown in Diagram 1, where TG1, T,, and TGzM are recalculated more precisely referring to the particular situation of the control cells in the experiments reported in Results.
Experimental Design for Drug-Induced Cell Cycle Perturbation Study Exponential growth is checked by cell count before starting drug treatment. PI/BrdUrd measurements are made according to the experimental design shown in Figure 2. Mean and standard errors at each point were obtained from three independent measurements, and three untreated flasks served as control. BrdUrd was added to the medium for 15 min before the end of treatment for pulse labeling. At the end of the treatment time, the cells of all flasks are washed and resuspended in drug- and BrdUrd-free medium for recovery. At the recovery times indicated in Figure 2, cells from the designated flasks are counted, centrifuged, and fixed. As can be seen in Results, the data a t -Ts/3 (2.4 h) give information about the output of G2M cells; the data a t -2Ts/3 (5.2 h) about the output from G1; and the data at -T,/3, -2Ts/3, and T, (8 h) taken together describe the progression through phase S
BrdUrd DNA ANALYSIS OF CELL CYCLE PERTURBATIONS
Distribution during exponential growth (t=Oh ) phase percentages whole cell number 29.5kO.8 58.3k1.4 G2M 12.2k1.2
Quiescent cells' phase percentages
whole cell number
121
phase cell number 295k8 583k15 122k12
phase cell number 23+10
2521k59 G2Mq
5+5
0.2k0.2
=(5)
Proliferating cells whole cell number 59.7k1.7 12.1k1.3
phase cell number
GF=0.965
117+13
TC=TD x lOg(1+GF) / log(2) TG2M=TC x log( 1+%G2M/100) / log(2) TG2M+TS=TC x lOg(1+(%S+%G2M)/lOO) / lOg(2)
TG1=2.55?0.63h TS= 7.20f0.46h
6
TC=lt.68+0.49h TG2=1.93+0.21 h TG2M+TS=9.13k0.40 h
I
DIAGRAM 1. Flow chart of the counts needed to calculate mean phase durations. "The number of quiescent cells a t time 0 is obtained from the percentage of BrdU negative cells in G1, S and G2M phases after continuous labeling up to 16 h (>TD), (see fig. 4) multiplied by the cell number at the same time. 'The number of quiescent cells in phases
G1, S and G2M are subtracted from the number of cells in phases G1, S, G2M a t time 0, to obtain the number of proliferating cells. '"The mean phase durations are calculated using the formulae 1-3, applied to the proliferating cells group. *The doubling time is calculated from Coulter counter cell number measurements.
(with qualitative discrimination between Searly-, Smiddle-, and Slate-cell behavior). Intervals of the order of Ts are suitable for reconstruction of the kinetics of treated cells at later times (e.g., -2Ts, -3Ts. . . 1, while avoiding subconfluency, where physiological alterations of the kinetics would constitute a n additional hurdle, even in the controls.
the possibility of inhibition of DNA synthesis during the treatment time was excluded. 2. The continuous labeling experiment shows up any cells made definitively quiescent during treatment, particularly the ones in G1 and G2M phases, not evaluable at step 1. All cells able to cross the S phase during the long (>T,) labeling time will incorporate BrdUrd, leaving only definitively quiescent cells in BrdUrd-negative compartments. This measure, however, does not provide information about the cells made quiescent after BrdUrd incorporation. The similarity of the proportions of proliferating cells in control and treated samples (Fig. 4) suggests that no cells were made unable to proliferate by the treatment (upper limit 1%). 3. At time t = 2.4 h (-Ts/3 and >T,,,), cells that were in Slateat the end of treatment have left the S phase and most cells in G2M are expected to have left GZM. This measure thus gives information about G2M
RESULTS Using the example of a particular dose of CNU, we describe the steps of the procedure for studying kinetic perturbations, leaving aside any pharmacological consideration, which is beyond the scope of this work. 1. From biparametric histograms obtained at 0 h (Fig. 3), it was possible to see whether the drug had made any cells unable to incorporate BrdUrd during pulse labeling. Since the percentages of BrdUrd-positive cells were the same for control and treated cells,
122 Pulse labeling:
c measuring times: 0 h
c 5.2h
2.4h
4 8h
16h
C o n t i n u o u s labeling:
t 16h
measuring times:
CNU treatment
a
BrdU exposure
FIG.2. Protocol of the study and measurement time for BrdUrd pulse and BrdUrd continuous labeling.
output. Since no BrdUrd-negative cells initially in G1 have had time to reach G2M, the cells in the G2M BrdUrd-negative window (Fig. 5, window C) are those that were present initially in G2M and have not yet completed this phase, although the mean G2 transit time is 2 h. Since the numbers of such cells were similar in control and treated samples, we conclude that the treatment did not alter the G2M output of cells initially in G2M phase.
CONTROL
4. The histograms of treated and control samples a t time t = 5.2 h (-2Ts/3), shown in Figure 6, show a n accumulation of BrdUrd-positive G2M cells. However, postponing analysis of these cells until the next time point, when their behavior became clearer, we consider here the output of cells initially in G1 phase (G1 output). This may be studied by calculating over time the number (not the percentage) of cells in the G1 BrdUrdnegative compartment, after subtraction of the contribution of the second cycle cells, estimated to be twice the G2M output (due to mitotic division) (Fig. 7). Data shown in Figure 7 indicate no alteration of Gloutput. 5 . At time t = 8 h (-Ts), most BrdUrd-positive cells (that were in S phase during labeling) have left S phase in control and many have reached the next (second) cycle, whereas BrdUrd-negative cells in either GI or in G2M phase initially are expected to be crossing the S phase. The procedure used to evaluate the number of cells that divided, in the group of cells in S phase a t the time of treatment, is a s follows: 1)calculate the number of their descendants (Fig. 8, window E; control: 0.574 x 1,445 = 829, treated: 378 x 1,289 = 378); 2) calculate the number of mother cells, assuming two descendants for each original S phase cell (control: 8291 2 = 414, treated: 378/2 = 189). The result is that 189 (46%)of the 414 BrdUrd-positive cells expected to divide within 8 h did so after treatment, whereas the remainder were delayed or blocked. In fact, Figure 8 shows a block in the G2M phase in treated samples. Moreover, similar analysis of the histograms for the previous time point showed that 86 (49%) of the 176 cells destined to reach the second cycle within 5.2 h (mainly cells initially in Slate)underwent mitosis in treated samples.
TREATED
60
30
DNA CONTENT FIG.3. Contour plot representations of biparametric flow cytometric histograms at 0 h recovery after BrdUrd pulse labeling. Control cell distribution (mean f s.e.1. Windows A: 29.5% f 0.4%,B: 1.0% f 0 . 2 % . C: 12.2% t 0 . 6 4 , D: 57.3% t 0.7%, cell number: 1,000 t 8.
*
Treated cell distribution: Windows A: 28.8% 1.4%,B: 0.9% f 0 . 2 4 , C : 11.8!% ! 0.6%, D: 5 8 . 5 4 i 1.5%, cell number: 1,000 f 40. The BrdlJrd content scale is linear.
BrdUrd DNA ANALYSIS OF CELL CYCLE PERTURBATIONS
CONTROL
123
TREATED
0
DNA CONTENT
30
60
DNA CONTENT
FIG.4. Contour plot representations of biparametric flow cytometric histograms at 16 h recovery with continuous exposure to BrdUrd. Control cell distribution (mean +- s.e.1: Window A: 0.9% t- 0.4%, B: 0.3% ? 0.2%,C: 0.2% i 0.2%, D: 98.6% f 0.5%,cell number 2,521 f 59. Treated cell distribution: Windows A: 1.3% t 0.2%, B: 0.4% 2 0.2%, C: 0.4% f 0.2%, D: 97.9% f 0.3%, cell number: 1,725 ? 118.
TREATED
CONTROL
I
ia
30
60
DNA CONTENT
30
60
DNA CONTENT
FIG.5. Contour plot representations of biparametric flow cytometric histograms at 2.4 h recovery after BrdUrd pulse labeling. Control cell distribution (mean i s.e.1: Windows A: 24.7% ? 1.2%,B: 15.2% +1.5%,C: 3.3% 2 0.5%, D: 53.7% 0.7%, E: 3.1% -+ 0.8%, cell number:
1,159 f 41. Treated cell distribution: Windows A: 23.1% 2 1.296, B: 14.4% ? 1.9%,C: 4.9% ? 0.6%, D: 55.7% ? 1.4%, E: 1.9% -t 0.2%, cell number: 1,103 ? 68.
Further information can be obtained about the fate of BrdUrd-positive cells over the S phase using the socalled relative movement (1,181, representing a DNA baricenter of the cloud of undivided cells. As is shown in Figure 9, at 5.2 h , treated samples showed slight delay in crossing S phase. Cells initially in Smiddle were delayed crossing Slatebut were not blocked; the relative movement value at t = 8 h was similar in treated and control samples. 6. At t = 16 h (-2T,), the control histogram (not shown) presented BrdUrd-positive cells in phases Slate
and G2M of the second cycle or in phases G1 and Searly of the third cycle, whereas BrdUrd-negative cells were spread over all cell cycle phases. However, in the treated sample, the picture was better interpreted by referring to the picture a t a previous time than by referring to the control for this time, since the time courses of controls and treated samples had moved too far apart at this time point. Comparing treated cell samples at t = 8 h and t = 16 h (Fig. lo), one can calculate first the number of BrdUrd-negative cells in G1 and S + G2M. Negative G1 cells are 79 ( = 0.061 x
*
124
UBEZIO ET AL.
TREATED
CONTROL
30
60
0
30
D N A CONTENT FIG.6. Contour plot representations of biparametric flow cytometric histograms a t 5.2 h recovery after BrdUrd pulse labeling. Control cell distribution (mean 2 s.e.1: Windows A: 14.5% 2 1.3%, B: 24.7% ? 2.0%, C: ND, D: 32.9'% 2 1.6%, E: 27.9% 2 0.770, cell number: 1,261
*
320
-
m
c
0
80 40
0 2
4
6
€3
time (h) 320
-lnm
0
c
--o-
&
120
f
80
P
z
-t-
Control Treated
40 0
48. Treated cell distribution: Windows A: 18.2% 2 0.7%, B: 23.2% ? 1.996, C: ND, D: 43.7% -t 1.8'76, E: 14.9% 2 2.296, cell number: 1,155 69.
*
16 h. Figure 10 (window BC) shows the presence of negative S phase cells at t = 8 h but virtually none a t t = 16 h . The interpretation of this observation is that BrdUrd-negative cells in S phase at t = 8 h reached G2M and were blocked there, together with the cells that had reached G2M previously, whereas cells in G1 phase a t t = 8 h remained blocked there. The total number of BrdUrd-positive cells increased, and pre317) - (378 + 402) = 360 cells divided cisely (823 and entered the second cycle, meaning that 360 2 154 (89%) of the 402 2 28 BrdUrd-positive cells, blocked at t = 8 h in the G2 phase of the first cycle, could overcome the block within the next 8-16 h interval. It is worth noting that the drug dose used in the experiment was sublethal. However, the use of cell numbers instead of percentages makes the method conceptually suitable for studying kinetic scenarios when cytotoxic effects are present. Cell loss a t different phases can be calculated, although cell loss cannot always be completely distinguished from kinetic movement.
+
0 0
60
DNA CONTENT
2
4
6
8
time (h)
FIG. 7. Time course of the number of G1 BrdUrd-negative cells without (a) and with (b) with subtraction of newborn cells in control and CNU treated samples.
1,289) a t t = 8 h and 88 (=0.051 x 1,725) a t t = 16 h. Negative S + G2M cells are 431 (=0.334 x 1,289) at t = 8 h and 496 (0.288 x 1,725) at t = 16 h. These data are compatible with a scenario in which the same cells are found in G1 and in S + G2M at t = 8 h and at t =
DISCUSSION This procedure optimizes experiments aimed at studying the kinetic perturbation produced by drug treatment in vitro. With BrdUrd pulse labeling during the last minutes of treatment and a continuous labeling measure, biparametric PI-BrdUrd analysis and cell count a t five time points after drug treatment are all that is needed to assess the fate of cells present initially in G1, S, and G2M phases of the cell cycle and to establish quantitatively the extent of their delay over a period corresponding to more than one doubling time of exponentially cycling control cells. However, because
BrdUrd DNA ANALYSIS OF CELL CYCLE PERTURBATIONS
CONTROL
0
125
TREATED
60
30
DNA CONTENT FIG.8. Contour plot representations of biparametric flow cytometric histograms a t 8 h recovery after BrdUrd pulse labeling. Control cell distribution (mean f S.E.): Windows A. 4.4% 2 0.5%, B,C: 26.0% f 0.7%, D: 12.2% t 0.6%, E: 57.4% f 0.7%,cell number: 1,445 f 38. Treated cell distribution: Windows A: 6.1% 2 0.9%, B,C: 33.4% f 1.6%, D: 31.2% t 1.4%, E: 29.3% ? 0.7%,cell number: 1,289 t 70.
t-
z W
3> 0
z
w
s ta
-1
w
a
0
2
4
6 a Time (h)
i
o
FIG. 9. Relative movement (mean 2 s.e.1 of control and CNU treated samples.
of the need for BrdUrd incorporation just before drug recovery, different strategies have to be adopted when drugs are used at concentrations inhibiting DNA synthesis. Other methods have been reported for analyzing BrdUrd DNA data in order to study quantitatively the kinetics of a cell population. Pallavicini et al. (9) derived G1, S,and G2M phase durations by fitting the data using a complex cell cycle model. This procedure requires a number of data points and specific software. With our approach we tried to obtain the maximum of kinetic information with a minimum of data points, using a pocket calculator. However, the need for information on the total cell number limits the use of our procedure to in vitro studies. Sasaki et al. (12) extracted data about the durations of S phase and of the whole cycle from the time course of the percentage of
labeled cells in the middle of S phase. In that approach, the durations of cell cycle phases are estimated by graphical procedures from data on the second cycle after treatment. Our method focuses on the flux of cells through each phase and on the percentages of blocked cells by the time of treatment. Phase durations in the control case can be assessed even with a procedure (Diagram 1)not requiring time course experiments. However, phase duration estimate is very difficult if the drug acts differently on cells in G1, S,and G2M phases during the treatment or if the mean time required to cross the subsequent phases is time dependent. For this reason, the data analysis we suggest here for treated cells is aimed at distinguishing drug effects on cells in phases G1, S, and G2M of the cell cycle, although the time required to complete the respective phases can be estimated using G2 and G1 output curves (Fig. 7) or the relative movement outline (Fig. 9). In the example reported, we found that L1210 cells in phases G1 and G2M at the end of the 1h treatment with 1 p,g/ml CNU were fully able t o leave the respective phases but were destined to remain blocked in the following G2M phase (G1 for a minority of them). We also found that cells initially present in S phase (BrdUrd positive after pulse labeling) had some minor delay in completing their S phase, and half of them remained temporarily blocked in the subsequent G2M phase. Such information could not be obtained by simple sequential DNA analysis, with which it is not possible to analyze separately the cells exposed to the drug in different phases, unless the cells are synchronized. Synchronization has its limitations: Incomplete synchrony or additive/synergistic effects of the synchronizing agent further add to the complexity of the experi-
126
UBEZIO ET AL.
TREATED 8 h RECOVERY
TREATED 16 h RECOVERY
i I-
z
8
3
em
0
30
60
DNA CONTENT
DNA CONTENT
FIG.10. Comparison of contour plots of CNU treated samples at 8 h and 16 h recovery after BrdUrd pulse labeling. Cell number distribution at 8 h (mean i s.e.): Windows A: 79 f 12 (6.1% f 0.9’36),B,C: 431 i 32 (33.4% i 1.6%),D: 402 i 28 (31.2% i- 1.4%),E: 378 i 22 (29.3%
i 0.7%). Cell number distribution at t = 16 h: Windows A: 88 t 16 (5.1% i 0.9%), B,C: 496 f 46 (28.8% i 1.8%), D: 823 & 62 (47.7% ? 1.6’361,E: 317 t 29 (18.4% -+ 1.1%).
mental design and require much more complex analysis than our method. In conclusion, our strategy permits the simultaneous analysis of the fate of cells exposed t o a drug in G1, S, and G2M phases, without additional perturbation of the system and with a very simple experimental design. The quantitative level attained in the proposed data analysis protocol makes this a useful tool for studying dose-kinetic responses of a drug and for comparing the kinetic effects of different drugs. The resulting information may be of great help in elucidating their mode of action.
7. Moran R, Darzynkiewicz 2, Staiano-Coico L, Melamed DR: Detection of 5-bromodeoxyuridine(BrdUrd) incorporation by monoclonal antibodies: Role of the denaturation step. J Histochem Cytochem 33:821-827, 1985. 8. Noguchi PD, Johnson JB, Browne W. Measurement of DNA synthesis by flow cytometry. Cytometry 1:390-393, 1981. 9. Pallavicini MG, Summers U,Dolbeare FD, Gray JW: Cytokinetic properties of asynchronous and cytosine arabinoside perturbed murine tumors measured by simultaneous bromodeoxyuridinelDNA analyses. Cytometry 6:602-610, 1985. 10. Pera F, Mattias P, Detzer K: Methods for determining the proliferation kinetics of cells by means of 5-Bromodeoxyuridine. Cell Tissue Kinet 10:255-264, 1977. 11. Sasaki K, Adachi S, Yamamoto T, Murakami T, Tonaka K, Takahashi M: Effects of denaturation with HCl on the immunological staining of bromodeoxyuridine incorporated into DNA. Cytometry 9:93-96, 1988. 12. Sasaki K, Murakami T, Ogino T, Takahashi M, Kawasaki S: Flow cytometric estimation of cell cycle parameters using a monoclonal antibody to bromodeoxyuridine. Cytometry 7:391-395, 1986. 13. Steel G G Growth Kinetics of Tumors. Cell Population Kinetics in Relation to the Growth and Treatment of Cancer. Clarendon Press, Oxford, 1977. 14. Stout RD, Suttles J : Problems and applications of cell cycle analysis: Distinguishing Go from G, and G, from S phase. Cytometry 3[Suppl.l:34-37, 1988. 15. Tice R, Schneider EL, Rary JM. The utilization of Bromodeoxyuridine incorporation into DNA for the analysis of cellular kinetics. Exp Cell Res 102:232-236, 1976. 16. Ubezio P, Rossotti A: Sensitivity of flow cytometric data to variations in cell cycle parameters. Cell Tissue Kinet 20507-517, 1987. 17. Vanderlaan M, Thomas CB: Characterization of monoclonal antibodies to bromodeoxyuridine. Cytometry 6:501-505, 1985. 18. White RA, Meistrich ML: A comment on “a method to measure the duration of DNA synthesis and the potential doubling time from a single sample.” Cytometry 7:486-490, 1986. 19. Vanagisawa M, Dolbeare F, Todoroki T, Gray JW: Cell cycle analysis using numerical simulation of bivariate DNNbromodeoxyuridine distributions. Cytometry 6:550-562, 1985.
ACKNOWLEDGMENTS The generous contribution of the Italian Association for Cancer Research, Milan, is gratefully acknowledged.
LITERATURE CITED 1. Begg AC, McNally NJ, Shrieve DC, Karcher H: A method to measure the duration of DNA synthesis and the potential doubling time from a single sample. Cytometry 6:620-626, 1985. 2. Dolbeare F, Beisker W, Pallavicini MG, Vanderlaan M, Gray JW: Cytochemistry for bromodeoxyuridineiDNA analysis: Stoichiometry and sensitivity. Cytometry 6521-530, 1985. 3. Dolbeare F, Gratzner H, Pallavicini MG, Gray JW: Flow cytometric measurement of total DNA content and incorporated bromodeoxyuridine. Proc Natl Acad Sci USA 8055734577, 1983. 4. Gratzner HG: Monoclonal antibody to 5-bromo- and 5-iododeoxyuridine: A new reagent for detection of DNA replication. Science 218:474-475, 1982. 5. Gray JW: Flow cytometry and cell kinetics: Relation to cancer therapy. In: Flow Cytometry IV, Laerum OD, Lindmo T, Thorud E (eds). Universitetsforlaget, Bergen, 1980, pp 485-491. 6. Miller MR, Heyneman C, Walker S, Ulrich R G Interaction of monoclonal antibodies directed against bromodeoxyuridine with pyrimidine bases, nucleosides, and DNA. J Immunol 136:17911795. 1986.
Cytometry 12:119-126 (1991)
Method for Kinetic Analysis of Drug-Induced Cell Cycle Perturbations Paolo Ubezio, Stefania Filippeschi, and Laura Spinelli Istituto di Ricerche Farmacologiche “Mario Negri,” 20157 Milano, Italy Received for publication May 21, 1990; accepted October 9, 1990
A method is described for quantitative study of the flux of cells through the cell cycle phases in in vitro systems perturbed by chemicals, such as chemotherapeutic agents. The method utilizes cell count and the flow cytometric technique of bromodeoxyuridine (BrdUrd)labeling, according to an optimized strategy. Cells are exposed to BrdUrd during the last minutes of drug treatment and fixed for analysis at 0,1/3T,, 2/3T,, T,, and T, + TG1 recovery times, where T,, TG1, T, are the mean durations of phases S and G1 and of the whole cycle of control cells. As an example of application of the proposed procedure, a kinetic study of the effect of l-(2-chloroethyl)-l-nitrosourea (CNU) on the L1210 cell cycle is described. Simple
Since the introduction in flow cytometry of the 5bromodeoxyuridine (BrdUrd) technique, the method has been recognized as a powerful tool for studying cell kinetics (3,8,10,14,15). However, the exact relationship between the flow cytometric (FCM) data obtained with this method and the true kinetic picture of a proliferating cell population is not simple. Only in a few studies have complex cell cycle models been used to fit FCM data (5,9,19). In most studies, only FCM data are presented, with no further attempt to obtain quantitative information through higher level kinetic data analysis. We devised a n optimized strategy that enabled us to obtain quantitative data on the flux of cells through cell cycle phases, using a minimum of FCM measurements with the propidium iodide (PI)/anti-BrdUrd method and simple data analysis. This analysis is approached in two steps. The first requires oriented analysis of biparametric DNA/antiBrdUrd distributions; this can be done as a “window” selection by all commercial computers fitted to flow cytometers. The second step requires only a pocket calculator to make the appropriate counts on first step FCM data. Cell numbers must be obtained by independent parallel measurements. The use of the method is illustrated by a n
data analysis, requiring only a pocket calculator, showed that cells in phases G1 and G2M at the end of a 1h treatment with 1pg/ml CNU were fully able to leave these phases but were destined to remain blocked in the following G2M phase (G1 for a minority of them). We also found that cells initially in S phase were slightly delayed in completing their S phase and that 50% of them remained temporarily blocked in the subsequent G2M phase, irrespective of their position in the S phase. Key terms: Antibromodeoxyuridine, flow cytometry, nitrosourea compounds, cell kinetics, dual-parameter analysis
experiment in which the cell cycle kinetic perturbation induced in vitro by a n anticancer drug is measured quantitatively, and the different effects in different phases are distinguished simultaneously.
MATERIALS AND METHODS Cell Culture L1210 murine leukemia cells were grown in RPMI 1640 medium (Gibco Europe, Glasgow, Scotland) supplemented with 20% heat-inactivated (56°C for 30 min) fetal calf serum (FCS; Flow Laboratories, Irvine, Scotland) as a suspension under standard culture conditions, at a density between 0.1 x lo6 and 0.5 x l o 6 celldm1 subcultured every 3 or 4 days. Drug Treatment l-(2-chloroethyl)-l-nitrosourea(CNU) was kindly provided by the Drug Synthesis and Chemistry Branch, Drug Therapeutics Program, National Cancer Institute (Bethesda, MD). Drug solutions in 95% ethanol were prepared immediately before use. The final ethanol concentration in medium after treatment never exceeded 0.01%.
UBEZIO ET AL.
120
500
0
6
12
Cell growth Before the kinetic study of drug-induced cell cycle perturbation, the kinetics of exponentially growing cells had to be studied to obtain information necessary for setting the experimental design. The in vitro growth curve was studied, and the best conditions of medium, serum, and seeding concentration were established to obtain reproducible balanced exponential growth over periods exceeding the duration of the drug experiment (about two doubling in our case). CNUlW/ml I
18
Time (h)
FIG.1. Growth of L1210 cells during experimental time. Control cells and cells treated with 1 pgiml CNU for 1 h before time 0. The growth curve data have been normalized to give the initial value of 1,000 cells/ml at time 0.
Label With BrdUrd Cells were seeded at 0.02 x lo6 cellsiml with 20% FCS and incubated for 48 h, then treated with 1 pg/ml of CNU for 1h at 37°C. The growth inhibition produced by the treatment is shown in Figure 1. In the last 15 min of treatment, cells were also incubated with 5 p M BrdUrd (Sigma). At the end of the incubation cells were washed in phosphate-buffered saline (PBS), pH 7.4, centrifuged, and resuspended in fresh medium. At the specific recovery times, cells were counted using a Coulter counter (Coulter Electronics Ltd., Harpenden, Herts, U.K.) equipped with a size analyzer (Coulter Channalyzer C-lOOO), fixed in 70% ethanol, and stored a t 4°C. PI-anti-BrdUrdStaining Technique The cells were stained by PI and anti-BrdUrd direct immunofluorescent method, slightly modifying the method of Dolbeare et al. (2). The ethanol-fixed cell suspensions were centrifuged and incubated with 3 N HC1 for 20 min, to obtain partially denatured DNA (7,ll). After washing with 0.1 M Na,B,O, to neutralize the acid, the cell pellets were resuspended with 50 pl of Tween 20 (Merck) 0.5% in PBS. After that, 50 p.1 of bovine serum albumin 0.5% in PBS and 20 pl of fluoresceinated anti-BrdUrd (4,6,17) (Becton Dickinson, Mountain View, CA) were added to each tube and incubated for 30 min a t room temperature. At the end of the antibody exposure, the cells were centrifuged, resuspended in 1 pg/ml of PI in PBS, and analyzed after a t least 1 h. Flow Cytometric Measurements A Facstar (Becton Dickinson) was used. Argon ion laser excitation was 100 mW at the 488 nm line. Green fluorescein fluorescence was detected in the 515-555 nm wavelength band and red PI fluorescence above 620 nm.
Basic Considerations The duration of cell cycle phases (TG1,T,, TGBM) during exponential growth was estimated from the percentage of cells in the cell cycle phases (%G1, %S, %G2M, which are constant in balanced growth), as detected by standard DNA-FCM methods and from the doubling time (TD),obtained by cell counts, using the following formulae (13): TCZM= Zog(2)-' x T D x log(1
(T,
+ %G2M/100)
(1)
+
TGBM) = Zog(2)r' x T D x Zog(1 + [%S + %G2Ml/100) T,,
=
TD - ( T , + T G 2 d .
(2)
(3)
Using these formulae, which are derived from the hypothesis of absence of quiescent or lost cells, and neglecting intercell variability, with our experimental model we obtained the following values: T, = 12 h, T,, = 2.5 h , T, = 7.5 h, and TGZM = 2 h. Previous work from our laboratory (16) showed that equations 1-3 hold with good approximation even in the presence of high intercell variability (CV of 50%), reinterpreting TG1, T,, and TGZM as mean values of the respective phase durations. In the presence of quiescent cells, we adopted the procedure shown in Diagram 1, where TG1, T,, and TGzM are recalculated more precisely referring to the particular situation of the control cells in the experiments reported in Results.
Experimental Design for Drug-Induced Cell Cycle Perturbation Study Exponential growth is checked by cell count before starting drug treatment. PI/BrdUrd measurements are made according to the experimental design shown in Figure 2. Mean and standard errors at each point were obtained from three independent measurements, and three untreated flasks served as control. BrdUrd was added to the medium for 15 min before the end of treatment for pulse labeling. At the end of the treatment time, the cells of all flasks are washed and resuspended in drug- and BrdUrd-free medium for recovery. At the recovery times indicated in Figure 2, cells from the designated flasks are counted, centrifuged, and fixed. As can be seen in Results, the data a t -Ts/3 (2.4 h) give information about the output of G2M cells; the data a t -2Ts/3 (5.2 h) about the output from G1; and the data at -T,/3, -2Ts/3, and T, (8 h) taken together describe the progression through phase S
BrdUrd DNA ANALYSIS OF CELL CYCLE PERTURBATIONS
Distribution during exponential growth (t=Oh ) phase percentages whole cell number 29.5kO.8 58.3k1.4 G2M 12.2k1.2
Quiescent cells' phase percentages
whole cell number
121
phase cell number 295k8 583k15 122k12
phase cell number 23+10
2521k59 G2Mq
5+5
0.2k0.2
=(5)
Proliferating cells whole cell number 59.7k1.7 12.1k1.3
phase cell number
GF=0.965
117+13
TC=TD x lOg(1+GF) / log(2) TG2M=TC x log( 1+%G2M/100) / log(2) TG2M+TS=TC x lOg(1+(%S+%G2M)/lOO) / lOg(2)
TG1=2.55?0.63h TS= 7.20f0.46h
6
TC=lt.68+0.49h TG2=1.93+0.21 h TG2M+TS=9.13k0.40 h
I
DIAGRAM 1. Flow chart of the counts needed to calculate mean phase durations. "The number of quiescent cells a t time 0 is obtained from the percentage of BrdU negative cells in G1, S and G2M phases after continuous labeling up to 16 h (>TD), (see fig. 4) multiplied by the cell number at the same time. 'The number of quiescent cells in phases
G1, S and G2M are subtracted from the number of cells in phases G1, S, G2M a t time 0, to obtain the number of proliferating cells. '"The mean phase durations are calculated using the formulae 1-3, applied to the proliferating cells group. *The doubling time is calculated from Coulter counter cell number measurements.
(with qualitative discrimination between Searly-, Smiddle-, and Slate-cell behavior). Intervals of the order of Ts are suitable for reconstruction of the kinetics of treated cells at later times (e.g., -2Ts, -3Ts. . . 1, while avoiding subconfluency, where physiological alterations of the kinetics would constitute a n additional hurdle, even in the controls.
the possibility of inhibition of DNA synthesis during the treatment time was excluded. 2. The continuous labeling experiment shows up any cells made definitively quiescent during treatment, particularly the ones in G1 and G2M phases, not evaluable at step 1. All cells able to cross the S phase during the long (>T,) labeling time will incorporate BrdUrd, leaving only definitively quiescent cells in BrdUrd-negative compartments. This measure, however, does not provide information about the cells made quiescent after BrdUrd incorporation. The similarity of the proportions of proliferating cells in control and treated samples (Fig. 4) suggests that no cells were made unable to proliferate by the treatment (upper limit 1%). 3. At time t = 2.4 h (-Ts/3 and >T,,,), cells that were in Slateat the end of treatment have left the S phase and most cells in G2M are expected to have left GZM. This measure thus gives information about G2M
RESULTS Using the example of a particular dose of CNU, we describe the steps of the procedure for studying kinetic perturbations, leaving aside any pharmacological consideration, which is beyond the scope of this work. 1. From biparametric histograms obtained at 0 h (Fig. 3), it was possible to see whether the drug had made any cells unable to incorporate BrdUrd during pulse labeling. Since the percentages of BrdUrd-positive cells were the same for control and treated cells,
122 Pulse labeling:
c measuring times: 0 h
c 5.2h
2.4h
4 8h
16h
C o n t i n u o u s labeling:
t 16h
measuring times:
CNU treatment
a
BrdU exposure
FIG.2. Protocol of the study and measurement time for BrdUrd pulse and BrdUrd continuous labeling.
output. Since no BrdUrd-negative cells initially in G1 have had time to reach G2M, the cells in the G2M BrdUrd-negative window (Fig. 5, window C) are those that were present initially in G2M and have not yet completed this phase, although the mean G2 transit time is 2 h. Since the numbers of such cells were similar in control and treated samples, we conclude that the treatment did not alter the G2M output of cells initially in G2M phase.
CONTROL
4. The histograms of treated and control samples a t time t = 5.2 h (-2Ts/3), shown in Figure 6, show a n accumulation of BrdUrd-positive G2M cells. However, postponing analysis of these cells until the next time point, when their behavior became clearer, we consider here the output of cells initially in G1 phase (G1 output). This may be studied by calculating over time the number (not the percentage) of cells in the G1 BrdUrdnegative compartment, after subtraction of the contribution of the second cycle cells, estimated to be twice the G2M output (due to mitotic division) (Fig. 7). Data shown in Figure 7 indicate no alteration of Gloutput. 5 . At time t = 8 h (-Ts), most BrdUrd-positive cells (that were in S phase during labeling) have left S phase in control and many have reached the next (second) cycle, whereas BrdUrd-negative cells in either GI or in G2M phase initially are expected to be crossing the S phase. The procedure used to evaluate the number of cells that divided, in the group of cells in S phase a t the time of treatment, is a s follows: 1)calculate the number of their descendants (Fig. 8, window E; control: 0.574 x 1,445 = 829, treated: 378 x 1,289 = 378); 2) calculate the number of mother cells, assuming two descendants for each original S phase cell (control: 8291 2 = 414, treated: 378/2 = 189). The result is that 189 (46%)of the 414 BrdUrd-positive cells expected to divide within 8 h did so after treatment, whereas the remainder were delayed or blocked. In fact, Figure 8 shows a block in the G2M phase in treated samples. Moreover, similar analysis of the histograms for the previous time point showed that 86 (49%) of the 176 cells destined to reach the second cycle within 5.2 h (mainly cells initially in Slate)underwent mitosis in treated samples.
TREATED
60
30
DNA CONTENT FIG.3. Contour plot representations of biparametric flow cytometric histograms at 0 h recovery after BrdUrd pulse labeling. Control cell distribution (mean f s.e.1. Windows A: 29.5% f 0.4%,B: 1.0% f 0 . 2 % . C: 12.2% t 0 . 6 4 , D: 57.3% t 0.7%, cell number: 1,000 t 8.
*
Treated cell distribution: Windows A: 28.8% 1.4%,B: 0.9% f 0 . 2 4 , C : 11.8!% ! 0.6%, D: 5 8 . 5 4 i 1.5%, cell number: 1,000 f 40. The BrdlJrd content scale is linear.
BrdUrd DNA ANALYSIS OF CELL CYCLE PERTURBATIONS
CONTROL
123
TREATED
0
DNA CONTENT
30
60
DNA CONTENT
FIG.4. Contour plot representations of biparametric flow cytometric histograms at 16 h recovery with continuous exposure to BrdUrd. Control cell distribution (mean +- s.e.1: Window A: 0.9% t- 0.4%, B: 0.3% ? 0.2%,C: 0.2% i 0.2%, D: 98.6% f 0.5%,cell number 2,521 f 59. Treated cell distribution: Windows A: 1.3% t 0.2%, B: 0.4% 2 0.2%, C: 0.4% f 0.2%, D: 97.9% f 0.3%, cell number: 1,725 ? 118.
TREATED
CONTROL
I
ia
30
60
DNA CONTENT
30
60
DNA CONTENT
FIG.5. Contour plot representations of biparametric flow cytometric histograms at 2.4 h recovery after BrdUrd pulse labeling. Control cell distribution (mean i s.e.1: Windows A: 24.7% ? 1.2%,B: 15.2% +1.5%,C: 3.3% 2 0.5%, D: 53.7% 0.7%, E: 3.1% -+ 0.8%, cell number:
1,159 f 41. Treated cell distribution: Windows A: 23.1% 2 1.296, B: 14.4% ? 1.9%,C: 4.9% ? 0.6%, D: 55.7% ? 1.4%, E: 1.9% -t 0.2%, cell number: 1,103 ? 68.
Further information can be obtained about the fate of BrdUrd-positive cells over the S phase using the socalled relative movement (1,181, representing a DNA baricenter of the cloud of undivided cells. As is shown in Figure 9, at 5.2 h , treated samples showed slight delay in crossing S phase. Cells initially in Smiddle were delayed crossing Slatebut were not blocked; the relative movement value at t = 8 h was similar in treated and control samples. 6. At t = 16 h (-2T,), the control histogram (not shown) presented BrdUrd-positive cells in phases Slate
and G2M of the second cycle or in phases G1 and Searly of the third cycle, whereas BrdUrd-negative cells were spread over all cell cycle phases. However, in the treated sample, the picture was better interpreted by referring to the picture a t a previous time than by referring to the control for this time, since the time courses of controls and treated samples had moved too far apart at this time point. Comparing treated cell samples at t = 8 h and t = 16 h (Fig. lo), one can calculate first the number of BrdUrd-negative cells in G1 and S + G2M. Negative G1 cells are 79 ( = 0.061 x
*
124
UBEZIO ET AL.
TREATED
CONTROL
30
60
0
30
D N A CONTENT FIG.6. Contour plot representations of biparametric flow cytometric histograms a t 5.2 h recovery after BrdUrd pulse labeling. Control cell distribution (mean 2 s.e.1: Windows A: 14.5% 2 1.3%, B: 24.7% ? 2.0%, C: ND, D: 32.9'% 2 1.6%, E: 27.9% 2 0.770, cell number: 1,261
*
320
-
m
c
0
80 40
0 2
4
6
€3
time (h) 320
-lnm
0
c
--o-
&
120
f
80
P
z
-t-
Control Treated
40 0
48. Treated cell distribution: Windows A: 18.2% 2 0.7%, B: 23.2% ? 1.996, C: ND, D: 43.7% -t 1.8'76, E: 14.9% 2 2.296, cell number: 1,155 69.
*
16 h. Figure 10 (window BC) shows the presence of negative S phase cells at t = 8 h but virtually none a t t = 16 h . The interpretation of this observation is that BrdUrd-negative cells in S phase at t = 8 h reached G2M and were blocked there, together with the cells that had reached G2M previously, whereas cells in G1 phase a t t = 8 h remained blocked there. The total number of BrdUrd-positive cells increased, and pre317) - (378 + 402) = 360 cells divided cisely (823 and entered the second cycle, meaning that 360 2 154 (89%) of the 402 2 28 BrdUrd-positive cells, blocked at t = 8 h in the G2 phase of the first cycle, could overcome the block within the next 8-16 h interval. It is worth noting that the drug dose used in the experiment was sublethal. However, the use of cell numbers instead of percentages makes the method conceptually suitable for studying kinetic scenarios when cytotoxic effects are present. Cell loss a t different phases can be calculated, although cell loss cannot always be completely distinguished from kinetic movement.
+
0 0
60
DNA CONTENT
2
4
6
8
time (h)
FIG. 7. Time course of the number of G1 BrdUrd-negative cells without (a) and with (b) with subtraction of newborn cells in control and CNU treated samples.
1,289) a t t = 8 h and 88 (=0.051 x 1,725) a t t = 16 h. Negative S + G2M cells are 431 (=0.334 x 1,289) at t = 8 h and 496 (0.288 x 1,725) at t = 16 h. These data are compatible with a scenario in which the same cells are found in G1 and in S + G2M at t = 8 h and at t =
DISCUSSION This procedure optimizes experiments aimed at studying the kinetic perturbation produced by drug treatment in vitro. With BrdUrd pulse labeling during the last minutes of treatment and a continuous labeling measure, biparametric PI-BrdUrd analysis and cell count a t five time points after drug treatment are all that is needed to assess the fate of cells present initially in G1, S, and G2M phases of the cell cycle and to establish quantitatively the extent of their delay over a period corresponding to more than one doubling time of exponentially cycling control cells. However, because
BrdUrd DNA ANALYSIS OF CELL CYCLE PERTURBATIONS
CONTROL
0
125
TREATED
60
30
DNA CONTENT FIG.8. Contour plot representations of biparametric flow cytometric histograms a t 8 h recovery after BrdUrd pulse labeling. Control cell distribution (mean f S.E.): Windows A. 4.4% 2 0.5%, B,C: 26.0% f 0.7%, D: 12.2% t 0.6%, E: 57.4% f 0.7%,cell number: 1,445 f 38. Treated cell distribution: Windows A: 6.1% 2 0.9%, B,C: 33.4% f 1.6%, D: 31.2% t 1.4%, E: 29.3% ? 0.7%,cell number: 1,289 t 70.
t-
z W
3> 0
z
w
s ta
-1
w
a
0
2
4
6 a Time (h)
i
o
FIG. 9. Relative movement (mean 2 s.e.1 of control and CNU treated samples.
of the need for BrdUrd incorporation just before drug recovery, different strategies have to be adopted when drugs are used at concentrations inhibiting DNA synthesis. Other methods have been reported for analyzing BrdUrd DNA data in order to study quantitatively the kinetics of a cell population. Pallavicini et al. (9) derived G1, S,and G2M phase durations by fitting the data using a complex cell cycle model. This procedure requires a number of data points and specific software. With our approach we tried to obtain the maximum of kinetic information with a minimum of data points, using a pocket calculator. However, the need for information on the total cell number limits the use of our procedure to in vitro studies. Sasaki et al. (12) extracted data about the durations of S phase and of the whole cycle from the time course of the percentage of
labeled cells in the middle of S phase. In that approach, the durations of cell cycle phases are estimated by graphical procedures from data on the second cycle after treatment. Our method focuses on the flux of cells through each phase and on the percentages of blocked cells by the time of treatment. Phase durations in the control case can be assessed even with a procedure (Diagram 1)not requiring time course experiments. However, phase duration estimate is very difficult if the drug acts differently on cells in G1, S,and G2M phases during the treatment or if the mean time required to cross the subsequent phases is time dependent. For this reason, the data analysis we suggest here for treated cells is aimed at distinguishing drug effects on cells in phases G1, S, and G2M of the cell cycle, although the time required to complete the respective phases can be estimated using G2 and G1 output curves (Fig. 7) or the relative movement outline (Fig. 9). In the example reported, we found that L1210 cells in phases G1 and G2M at the end of the 1h treatment with 1 p,g/ml CNU were fully able t o leave the respective phases but were destined to remain blocked in the following G2M phase (G1 for a minority of them). We also found that cells initially present in S phase (BrdUrd positive after pulse labeling) had some minor delay in completing their S phase, and half of them remained temporarily blocked in the subsequent G2M phase. Such information could not be obtained by simple sequential DNA analysis, with which it is not possible to analyze separately the cells exposed to the drug in different phases, unless the cells are synchronized. Synchronization has its limitations: Incomplete synchrony or additive/synergistic effects of the synchronizing agent further add to the complexity of the experi-
126
UBEZIO ET AL.
TREATED 8 h RECOVERY
TREATED 16 h RECOVERY
i I-
z
8
3
em
0
30
60
DNA CONTENT
DNA CONTENT
FIG.10. Comparison of contour plots of CNU treated samples at 8 h and 16 h recovery after BrdUrd pulse labeling. Cell number distribution at 8 h (mean i s.e.): Windows A: 79 f 12 (6.1% f 0.9’36),B,C: 431 i 32 (33.4% i 1.6%),D: 402 i 28 (31.2% i- 1.4%),E: 378 i 22 (29.3%
i 0.7%). Cell number distribution at t = 16 h: Windows A: 88 t 16 (5.1% i 0.9%), B,C: 496 f 46 (28.8% i 1.8%), D: 823 & 62 (47.7% ? 1.6’361,E: 317 t 29 (18.4% -+ 1.1%).
mental design and require much more complex analysis than our method. In conclusion, our strategy permits the simultaneous analysis of the fate of cells exposed t o a drug in G1, S, and G2M phases, without additional perturbation of the system and with a very simple experimental design. The quantitative level attained in the proposed data analysis protocol makes this a useful tool for studying dose-kinetic responses of a drug and for comparing the kinetic effects of different drugs. The resulting information may be of great help in elucidating their mode of action.
7. Moran R, Darzynkiewicz 2, Staiano-Coico L, Melamed DR: Detection of 5-bromodeoxyuridine(BrdUrd) incorporation by monoclonal antibodies: Role of the denaturation step. J Histochem Cytochem 33:821-827, 1985. 8. Noguchi PD, Johnson JB, Browne W. Measurement of DNA synthesis by flow cytometry. Cytometry 1:390-393, 1981. 9. Pallavicini MG, Summers U,Dolbeare FD, Gray JW: Cytokinetic properties of asynchronous and cytosine arabinoside perturbed murine tumors measured by simultaneous bromodeoxyuridinelDNA analyses. Cytometry 6:602-610, 1985. 10. Pera F, Mattias P, Detzer K: Methods for determining the proliferation kinetics of cells by means of 5-Bromodeoxyuridine. Cell Tissue Kinet 10:255-264, 1977. 11. Sasaki K, Adachi S, Yamamoto T, Murakami T, Tonaka K, Takahashi M: Effects of denaturation with HCl on the immunological staining of bromodeoxyuridine incorporated into DNA. Cytometry 9:93-96, 1988. 12. Sasaki K, Murakami T, Ogino T, Takahashi M, Kawasaki S: Flow cytometric estimation of cell cycle parameters using a monoclonal antibody to bromodeoxyuridine. Cytometry 7:391-395, 1986. 13. Steel G G Growth Kinetics of Tumors. Cell Population Kinetics in Relation to the Growth and Treatment of Cancer. Clarendon Press, Oxford, 1977. 14. Stout RD, Suttles J : Problems and applications of cell cycle analysis: Distinguishing Go from G, and G, from S phase. Cytometry 3[Suppl.l:34-37, 1988. 15. Tice R, Schneider EL, Rary JM. The utilization of Bromodeoxyuridine incorporation into DNA for the analysis of cellular kinetics. Exp Cell Res 102:232-236, 1976. 16. Ubezio P, Rossotti A: Sensitivity of flow cytometric data to variations in cell cycle parameters. Cell Tissue Kinet 20507-517, 1987. 17. Vanderlaan M, Thomas CB: Characterization of monoclonal antibodies to bromodeoxyuridine. Cytometry 6:501-505, 1985. 18. White RA, Meistrich ML: A comment on “a method to measure the duration of DNA synthesis and the potential doubling time from a single sample.” Cytometry 7:486-490, 1986. 19. Vanagisawa M, Dolbeare F, Todoroki T, Gray JW: Cell cycle analysis using numerical simulation of bivariate DNNbromodeoxyuridine distributions. Cytometry 6:550-562, 1985.
ACKNOWLEDGMENTS The generous contribution of the Italian Association for Cancer Research, Milan, is gratefully acknowledged.
LITERATURE CITED 1. Begg AC, McNally NJ, Shrieve DC, Karcher H: A method to measure the duration of DNA synthesis and the potential doubling time from a single sample. Cytometry 6:620-626, 1985. 2. Dolbeare F, Beisker W, Pallavicini MG, Vanderlaan M, Gray JW: Cytochemistry for bromodeoxyuridineiDNA analysis: Stoichiometry and sensitivity. Cytometry 6521-530, 1985. 3. Dolbeare F, Gratzner H, Pallavicini MG, Gray JW: Flow cytometric measurement of total DNA content and incorporated bromodeoxyuridine. Proc Natl Acad Sci USA 8055734577, 1983. 4. Gratzner HG: Monoclonal antibody to 5-bromo- and 5-iododeoxyuridine: A new reagent for detection of DNA replication. Science 218:474-475, 1982. 5. Gray JW: Flow cytometry and cell kinetics: Relation to cancer therapy. In: Flow Cytometry IV, Laerum OD, Lindmo T, Thorud E (eds). Universitetsforlaget, Bergen, 1980, pp 485-491. 6. Miller MR, Heyneman C, Walker S, Ulrich R G Interaction of monoclonal antibodies directed against bromodeoxyuridine with pyrimidine bases, nucleosides, and DNA. J Immunol 136:17911795. 1986.
