0 1991 Wiley-Liss, Inc.

t

Discrepancies Among the Metaphase, Telophase, and the G,/G, and G, DNA Peaks of Heteroploid Cell Lines William C. Dooley,’ David C. Allison: and Joel Robertson Department of Surgery, The Johns Hopkins Medical Institutions, Baltimore, Maryland 21205 (W.C.D., D.C.A.), Department of Surgery, Baltimore VA Medical Center, Baltimore, Maryland 21218 (D.C.A.), and Robertson Electronics, Albuquerque, New Mexico 87123 (J.R.) Received for publication April 10, 1990; accepted October 17, 1990

Heteroploid cell populations often show narrow peaks of GdG, and G,/M DNA content and broadly distributed chromosome numbers. This was originally explained by the selective metaphase arrest of the cells that have nonmodal chromosome numbers. To test whether this explanation applies, we have measured the chromosome number distributions, as well as the G,/G1, G,, metaphase (M), and telophase (T) DNA distributions, of the cell lines WCHE-5, MCa-11, and HL-60. The WCHE-5 cells had narrowly distributed chromosome numbers and G,/G, G,, M, and T DNA peaks. The MCa-11 and HL-60 cells also had narrowly distributed G,/G, and G,

Karyotype analyses of individual metaphase cells within a given tumor or cultured cell line often reveal a wide variation in chromosome form and number (10,14,12). Surprisingly, measurements of the interphase DNA distributions of such chromosomally diverse cell populations often reveal narrow G,/G, and G,/M DNA peaks (11,16). This discrepancy between the narrow interphase Go/ G, and G,/M DNA peaks and the widely dispersed chromosome numbers of heteroploid cell populations has been used in support of certain aspects of the “stem cell theory” of cancer (14). In a n early study, Stich found that the DNA contents of the metaphase cells of aneuploid tumors were widely distributed, whereas the DNA contents of the post-karyokinetic daughter nuclei of telophase or anaphase cells in the same tumors were narrowly distributed (18,19). These data seemed to show that only cells of the “main stemline,” or those cells with DNA contents and chromosome numbers a t or near the modal value, were capable of completing mitosis and re-entering interphase for another replicative cycle. Thus, it was believed that the metaphase cells with non-modal DNA contents and chromosome

DNA peaks, but broadly distributed chromosome numbers and M and T DNA peaks. The widths of the MCa-11 and HL60 M- and T-cell DNA peaks were similar to those of their chromosome number peaks, suggesting that all cells were completing mitosis, regardless of chromosome number or DNA content. Thus, selective metaphase arrest does not seem to be the cause of the narrow GdG, and G, DNA peaks of heteroploid cell populations. Key terms: DNA content, stem cell theory, chromosome number distributions, absorption cytometry, heteroploid cell lines, aneuploidy

numbers were arrested a t mitosis and therefore were not represented in the post-karyokinetic, telophase DNA distributions. These findings satisfactorily explained the discrepancies between the widths of the interphase Go/G1and G, DNA peaks and the chromosome number distributions often observed for heteroploid tumors and cell lines. Specifically, the postulated selective metaphase arrest of the non-modal cells led to a relative overrepresentation of these non-modal cells in the chromosome number distributions obtained from analysis of standard metaphase karyotype spreads. The narrow G,/G, and G, peaks of the interphase DNA distributions were thus, in turn, almost solely representative of cells from the immortal “main stemlines.” Such “stemlines” were thought to be relatively homogeneous in terms of chromosome numbers as well as G,/G, and G2 DNA content. ‘Address reprint requests to William C. Dooley, M.D., Carnegie 681, The Johns Hopkins Hospital, 600 N. Wolfe Street, Baltimore, MD 21205. ‘Supported by the Department of Veteran Affairs.

100

DOOLEY E T AL.

In 1971, Kraemer and associates at Los Alamos reported the results of a series of experiments involving flow cytometric measurements of the interphase DNA contents of several heteroploid cell lines (11).The DNA contents of these cell lines were measured both with and without colcemid treatment. They found that the widths of the G,/G, and G,/M peaks, as measured by the coefficients of variation (CV, s/x x 100) of the DNA measurements of these peaks, were identical and uniformly narrow regardless of whether or not colcemid treatment was employed. These workers interpreted their results a s providing evidence against the existence of a large proportion of “metaphase-arrested” cells with non-modal DNA content in these cell lines. This discrepancy between diverse chromosome numbers in the metaphase cells of heteroploid lines and their narrow DNA peaks of G,/G, and G2/M cells has come to be known a s the “Los Alamos Paradox.” As a possible solution to this paradox, Kraemer et al. suggested that the metaphase cells with diverse chromosome numbers within a given heteroploid cell population might actually have the same amounts of DNA per cell. We have re-examined this question by making standard karyotypic chromosome counts and absorption cytometric measurements of the DNA contents of the metaphase, telophase, Go/G1, and G2 cells of one relatively chromosomally stable (WCHE-5) and two heterogenous ploidy (MCa-11 and HL-60) cell lines. The results of this study were surprising, in that neither the model of classic “stem cell theory” nor that of Kraemer et al. seemed to explain the presence of narrow G,/G, and G2 DNA peaks in chromosomally diverse cell populations.

slides and flame drying. Standard chromosome counts were performed after standard Giemsa staining. Chicken erythrocytes were added to the second portion of the cells harvested from each flask as a n internal standard for the DNA measurements. These cells were then fixed with modified Carnoy’s without the hypotonic treatment. Slides were prepared in a similar manner as for the chromosome preparations. No significant differences were found in the chromosome number and DNA distributions obtained from replicate cultures in these experiments.

Measurements of DNA Content of Mitotic and Interphase Cells After slide preparation, the cells upon which the absorption cytometric DNA measurements were to be made were initially stained by a modification of the Giemsa staining procedure which makes possible the rapid identification of mitotic cells (6). The Giemsastained cells were classified by light microscopy as being either in interphase, or in the pre-karyokinetic (prophase and metaphase) or post-karyokinetic (anaphase and telophase) phases of mitosis. This information, as well as the slide position of each cell, recorded with a 0.5 km computerized microscope stage, was automatically stored in a personal computer file by use of a previously described hardwarelsoftware control system (Robertson Electronics, Albuquerque, NM). After the cells had been classified and their slide position mapped, the Giemsa stain was removed with acid alcohol. The cells were then Feulgen stained by hydrolysis in 4 N HC1 at 28°C for 1h, followed by incubation in Schiff’s reagent (Sigma, St. Louis, MO) for 1 h at room temperature. After Feulgen staining, the previously mapped and MATERIALS AND METHODS morphologically classified cells were automatically reCell Culture and Slide Preparation found with the computerized stage. Absorption cytomThe MCa-11 (a mouse mammary carcinoma line) and etry of cellular DNA contents was performed with a WCHE-5 (a Chinese hamster embryo line) cells were Vickers M85 microdensitometer with a 100 x , 1.25 NA grown in exponential monolayer cultures for three gen- achromatic lens. A 0.4krn spot was used for the absorerations from a freezer stock. The MCa-11 cells were bance measurements. In its scanning mode, this instrugrown in MEM medium with 10% fetal calf serum ment makes up to 50,000 point-absorbance measure(FCS) supplemented with penicillin and streptomycin. ments over a selected slide area that is defined by a The WCHE-5 cells were grown in alpha-Earles me- mask placed over each nucleus to be measured. The dium with 20% FCS supplemented with penicillin and sum of the point absorbencies measured for a given streptomicin. A harvest of mitotic cells was performed nucleus is proportional to its integrated absorbance, when the monolayer cells had grown to between two- and hence to its total Feulgen stain content. Absorthirds and three-quarters confluency. HL-60, a human bance measurements were made a t light wavelengths promyelocytic leukemia cell line, was grown in sus- selected between 590-615 nm, with spectral bandpension cultures in RPMI-1640 medium with 10% widths of 15 nm and a 4% instrument glare level, FCS supplemented with penicillin and streptomycin. which minimize optical measuring errors. The inteTwenty-four hours after the last split, all of the cells in grated absorbencies of the chicken erythrocytes on the suspension cultures were harvested from the each slide were used as internal standards for cellular DNA content (3,4). flasks. After harvest, the cells from each of the three lines The mitotic cells were classified on the basis of their were divided into two portions. The first portion under- morphology as being either pre-karyokinetic (M) or went standard hypotonic treatment, followed by fixa- post-karyokinetic (T). More than 75% of the M cells tion in modified Carnoy’s. Chromosome spreads were were in metaphase. The DNA distributions of the prepared from these fixed cells by dropping them onto prophase and metaphase cells did not differ signifi-

DNA CONTENT OF HETEROPLOID CELLS

101

of

cells

cells


80 Chromosome

2 0 40

60

cells


80 Chromosome

2 0 40

60




80

Chromosome a

FIG.1. Chromosome number histograms from non-colcemid-treated exponentially growing cells of each of three cell lines. Greater than 200 metaphase spreads were counted in each case. (A) WCHE-5; (B) MCa-11; (C)HL-60.

cantly (data not shown). The majority of T cells were either in late anaphase or early telophase, and, again, no differences were found in the DNA distributions of these two cell populations (data not shown). The DNA intervals for calculation of the CVs of the Go/Gl and G, peaks of the interphase DNA distributions were selected a s follows: the lower range of DNA content for each peak was chosen between the point where the inflection of the slope turned upward, and the upper range of DNA content was defined a t the inflection point of the downward slope. All DNA values between these boundaries were used to calculate the CVs for each G,/G, and G, DNA peak. This is therefore a very inclusive method of determining CVs which attempts to encompass the whole peak. The “112 peak width method”, which decreases the CV, was not used. The chromosome numbers for all cells measured in the chromosome counts, and the DNA values of all M and T cells upon which DNA measurements were performed, were used for calculation of the respective CVs of the chromosome numbers, and of the M and T cell DNA distributions (1,2).

RESULTS In order to learn more about the mechanisms responsible for the apparent narrowness of the G,/G, and G2 DNA peaks in heteroploid cell lines, we measured the chromosome numbers and the interphase-cell, M-cell (prophase and metaphase cells), and T-cell (telophase and anaphase cells) DNA distributions in the WCHE-5, MCa-11, and HL-60 cell lines. The WCHE-5 cells showed a very narrow distribution of chromosome numbers under the culture conditions used in these experiments (Fig. 1A). Most of the karyotype variation seen for this cell line involved trisomy alternating with tetrasomy for one of the smallest chromosomes. The MCa-11 line, originally derived from a mouse mammary carcinoma, showed a broad distribution of chromosome numbers in monolayer culture (Fig. 1B). In the

HL-60 line, derived from a human patient with promyelocytic leukemia, double minute chromosomes can occasionally form. In the experiments reported, however, the HL-60 cells were found to have centromeric chromosomes exclusively. The chromosome numbers of the HL-60 cells (Fig. 1C) were broadly distributed (Fig. lC), a s were those of the MCa-11 cells (Fig. 1B). The CVs of the WCHE-5, MCa-11, and HL-60 chromosome number distributions were 2,27, and 16%, respectively (Table 1). Figure 2 shows the interphase DNA content distributions for the WCHE-5 (Fig. 2A), MCa-11 (Fig. 2B), and HL-60 (Fig. 2C) lines. The G,/G, and G, DNA peaks for each of these lines were relatively narrow and appeared not to differ significantly from one another. All of the CVs of the interphase G,/G, and G, peaks for each of the lines were 7.9% or less (Table 1). Thus, the MCa-11 and HL-60 lines seemed to have relative narrowness of the Go/G1 and G2 DNA peaks, despite the fact that the chromosome numbers of these lines were broadly distributed (Figs. 1 , 2 , Table 11, confirming the observation of Kraemer and others of relatively narrow G,/G, and G, DNA peaks in certain heteroploid cell lines (11). The DNA distributions of the M cells (metaphase and prophase cells) for each of the lines are illustrated in Figure 3. Figure 4 shows the T-cell (anaphase and telophase cells) DNA distributions for these lines. As expected, the WCHE-5 line showed relatively narrow peaks of DNA content for both its M (Fig. 3A) and T cells (Fig. 4A). The CVs of the WCHE-5 M- and T-cell DNA measurements corresponded very closely to the CVs of the G,/G1 and G, DNA peaks (Fig. 2A, Table 1). The MCa-11 and HL-60 lines, however, showed a much greater degree of variation in their M and T DNA contents (Figs. 3B, 3C, 4B, 4C) than did the WCHE-5 line (Fig. 3A, 4A). For both the MCa-11 and HL-60 lines, the mitotic-cell DNA content peak was wider than its corresponding interphase counterpart. Specif-

102

DOOLEY ET AL.

Table 1 Comparison of Peak Width by using Coefficient of Variation cvs

GI

3M Chr #

WCHE-5 6.9 6.7 7.3 9.3 2.4

MCa-11 6.4 6.6 16.5 20.5 27.3

HL-60 6.1 7.9 11.8 13.4 16.2

ically, the MCa-11 and HL-60 T-cells (Fig. 4B, C) had wider DNA peaks than did their respective G,/G, cells (Fig. 2B, C). Similarly, the MCa-11 and HL-60 M cells had wider DNA peaks (Fig. 3B, C) than did their respective G, cells (Fig. 2B, C). The CVs of these peaks are listed and compared in Table 1. For both MCa-11 and HL-60 lines, the CVs of the T-cell DNA measurements were slightly lower than the CVs of the M-cell DNA measurements, but were higher than the CVs of their respective G,/G, DNA measurements (Table 1). Thus, the MCa-11 and HL-60 lines present a paradox of narrow interphase G,/G, and G, DNA content peaks (Fig. 2B, C) in the face of varying chromosome numbers (Fig. 1) and widely dispersed M and T DNA contents (Table 1). The increased dispersion of the mitotic cell DNA measurements of the HL-60 and MCa-11 line, compared to the relatively narrow dispersion of the mitotic cell DNA measurements of the WCHE-5 line and the interphase DNA measurements of all three lines, can be illustrated by a comparison of the ratios of the CVs of these DNA measurements (Table 1). The ratios of the CVs of the G,/G, to G, cell DNA measurements were 1.02, 0.77, and 0.97 for the WCHE-5, HL-60 and MCa-11 Iines, respectively. The corresponding ratios of the M- and T-cell DNA measurement CVs were 1.28, 1.14, and 1.25 for these three lines. Marked differences were seen, however, between the WCHE-5, as opposed to the MCa-11 and HL-60 lines, when the CV ratios of their mitotic DNA measurements to their corresponding interphase DNA measurements were compared. For the WCHE-5 line, the CV ratios of the T to G,/G, and the M to G, DNA measurements were 1.05 and 1.38, respectively. These CV ratio values are similar to the values for the CV ratios of the interphase DNA peaks and the mitotic DNA peaks of all three lines (see above). The CV ratios of the T to G,/G, and the M to G, DNA measurements were, however, markedly elevated for the HL-60 and MCa-11 lines, being 1.90 and 1.70, and 2.58 and 3.11, respectively. Thus, the G,/G,, G,, and M and T DNA measurements of the WCHE-5 lines show reasonably similar, and low, degrees of dispersion (Table 1).The M and T DNA measurements of the HL60 and MCa-11 lines show relatively greater dispersion than the DNA measurements of their corresponding interphase peaks (Table l ) , but the relative dispersions of these M- and T-cell DNA measurements are similar within each of these lines (and most likely within the range of measurement errors of the cytophotometric system).

Although it is obvious that, due to variation in individual chromosome size and DNA content, a rigorous correlation between chromosome number and metaphase DNA content would not be expected, it is still of interest to note that a rough correlation between these parameters was found for the WCHE-5, HL-60, and MCa-11 lines. Specifically, if one takes the chromosome number ranges which encompass 90% of the karyotypes counted for the WCHE-5, HL-60, and MCa11lines, these numbers would be 20 2 1,35 2 5, and 59 2 40 chromosomes, respectively (Fig. 1). The corresponding CVs of the metaphase DNA measurements for each of these lines were 9.3, 13.4, and 20.5%, respectively (Fig. 2, Table 1). Thus, even though there are differences in individual chromosomal DNA content, these findings clearly show that variations in metaphase chromosome number are reflected by variations in metaphase DNA content.

DISCUSSION Stich observed in 1962 that the cells of certain solid tumors seemed to have widely distributed M DNA contents and relatively narrowly distributed T DNA values (18). These findings were interpreted as showing that many tumor M cells with DNA contents away from the main stemline were “metaphase arrested” and could not complete mitosis. This observation provided major support for certain aspects of the “stem cell theory” of cancer (9,10,13,14). This theory postulates that the chromosomal diversity of many heteroploid tumors is due, in large part, to the selective metaphase arrest of cells with non-modal chromosome numbers. Furthermore, the cells of the “main stemline,” which have relatively homogeneous chromosome complements and DNA contents, complete mitosis rapidly and make up the majority of the cycling cells found in the interphase DNA distribution. Although attractive in explaining the disparity between the widely spread chromosome numbers and tight G,/G, and G, DNA peaks observed for many tumors and cell lines, this version of the stem cell theory does pose some difficulties. Direct microscopic inspection of the sections from most solid tumors does not usually reveal large proportions of “metaphasearrested’ cells. Also, if only cells with the modal chromosome number and DNA content complete mitosis, it is not clear why the non-modal M cells would make up such a large proportion of the chromosome number distributions of these heteroploid cell populations. For example, in this study, the MCa-11 and HL-60 cells with non-modal chromosome numbers, a s well as non-modal M cell DNA contents, made up approximately 50% of the total population of cells measured in the chromosome number and M DNA content distributions (Figs. 1 and 3). It can be argued that, even if such non-modal cells were immortal, such cells would still become increasingly rare in the total cell population because of their inability to replicate. This is clearly not the case, however, as these non-modal cells are abundant (Figs.

103

DNA CONTENT OF HETEROPLOID CELLS

INTERPHASE 25

25

A

25

B

B 20

%

20

c

20

15

Of

15

Of

15

Of

cells

cells

C

L

cells

10

10

5

5

10

C

DNA(in pg.)

DNA(in pg.)

DNA(in pg.)

FIG.2. Interphase DNA histograms obtained from exponentially growing cultures of each of three cell lines. DNA content was determined by absorption cytometry of Feulgen-stained cells with internal controls of chicken erythrocytes. Histograms were developed on measurements of greater than 500 interphase cells. (A) WCHE-5; (B) MCa-11; (C) HL-60.

METAPHASE 25

%

25

A

20

%

of

2:

B

% 2c

20

of

of 15

1!

15

cells

cells

cells

.-

in

10

5

II

0 0

10

1C

5A

30

DNA(in pg.)

iL

C

0 20

C

10

0

20

30

DNA(in pg.)

0

10

20

30

DNA(in pg.)

FIG.3. DNA content histograms of morphologically identified metaphase cells in each of three cell lines. DNA content was determined by absorption cytometry of previously mapped metaphase cells (>300)after Feulgen staining with the use of chicken erythrocytes as internal controls. (A)WCHE-5; (B) MCa-11; (C) HL-60.

1 and 3). Finally, direct time lapse photographic studies of cultured heteroploid cell lines have shown that almost all M cells with differing chromosome numbers actually do complete mitosis, albeit a t different rates (9,15,17). Given such questions about this aspect of the stem cell theory, we have repeated the basic portions of Stich's original experiments with, what we hope, are several technical improvements. In the present work, the relationships between heteroploidy and G,/G,, G,, M, and T DNA contents have been determined on slides prepared from single cell suspensions of cultured cells

rather than on intact tumor sections. This avoids the difficulty of performing DNA measurements on sections that contain overlapping cells. The M and T cells in this study were identified, and had their slide positions mapped, by specific staining prior to the performance of DNA measurements. This specific staining procedure allowed eventual performance of DNA measurements on relatively large numbers of the M and T cells, and it minimized unintentional observer bias in the selection of M and T cells for DNA measurements. The semi-automated cytometric system allowed many absorption cytometric measurements of interphase

104

DOOLEY ET AL.

TELOPHASE 40

40

A

30

30

30 of

of

of

cells

20

cells

20

20

10

10

0

C

%

%

%

cells

40

B

Y

L 10

0

DNA(in pg.)

DNA(in pg.)

20

30

DNA(in pg.)

FIG.4. DNA content histograms of morphologically identified telophase cells. Data were obtained by absorption cytometry of Feulgen-stained morphologically identified telophase cells (1300) in each of three cell lines. (A) WCHE-5; (B) MCa-11; (C) HL-60

cells to be made from which we were able to obtain valid interphase DNA distributions. Finally, considerable attention was paid to the reduction of optical errors in the measurement of DNA content by absorption cytometry. This was done as follows: The light wavelengths employed for the DNA measurements were selected as far off peak as possible to lower the stain darkness of the mitotic cells and thereby minimize diffraction, residual distributional, and glare errors (3,4). A computerized “look up table” was used to correct all absorbances for optical errors due to stain darkness (3). The average of the integrated absorbancelnuclear area ratios, a parameter that directly correlates with optical density (3), was virtually identical for the mitotic cells for all three lines. In addition, there was very little variation in the optical densities of the mitotic cells each of the three lines; the CVs of the integrated absorbancelnuclear area ratios of all mitotic cells measured in each of the lines only ranged between 2.3 and 2.8%. Finally, the M-cells upon which DNA measurements were performed were prepared without hypotonic treatment to minimize heterogeneous staining and chromosomal edges which would increase residual distributional and diffraction errors. The mitotic cells of all three lines appeared morphologically similar after this method of preparation. In spite of these precautions, there still might be more variation due to optical errors in the mitotic than the interphase DNA measurements. Thus, while one would expect all of the DNA measurements of the WCHE-5 line to show identical variation, the CVs of the DNA measurements of the mitotic WCHE-5 cells are slightly higher than the DNA measurements per-

formed on the corresponding interphase cell populations: a n M CV of 9.1 versus a G, CV of 6.7%, and a T CV of 7.3 compared to a GolGl CV of 6.9% were found for the WCHE-5 line (Table 1).Nevertheless, even if these apparent differences are really significant and actually due to increased measuring errors for the mitotic WCHE-5 cells, the magnitude of these possible measuring errors seems to be small, and nowhere near the size of differences observed for the CVs of the interphase- and mitotic-cell DNA measurement of the HL-60 and MCa-11 lines (Table 1).Thus, we feel i t is unlikely (although not absolutely excluded) that systematic optical measuring errors are the explanation for the overall differences found for the mitotic and interphase DNA measurements of the WCHE-5, HL60, and MCa-11 lines. Thus, the results obtained for the WCHE-5 line would seem to validate the accuracy of this cytometry system. Specifically, the WCHE-5 cells were found to have narrow peaks of chromosome numbers (Fig. 1A) and of G,/G,, G, (Fig. 2A), M (Fig. 3A), and T (Fig. 4A) DNA content (Table 1).These results are expected for DNA measurements performed on a line with little cell to cell variation in its chromosome composition. The results of the DNA measurements performed on the MCa-11 and HL-60 lines, however, were somewhat surprising. We found, as the stem cell theory predicts, that the dispersion of the DNA content measurements performed on the heteroploid MCa-11 and HL-60 M cells (Fig. 3B, C) closely paralleled the dispersion of the chromosome number distributions of the two lines (Fig. lB, C, Table 1). These observations are not consistent with the hypothesis of Kraemer et al., who postulated that the chromosomally diverse M cells present in such

DNA CONTENT OF HETEROI’LOID CELLS

heteroploid lines may have roughly identical DNA contents (11).Contrary to classic stem cell theory, however, we found that the degree of dispersion of the DNA measurements performed on MCa-11 and HL-60 Tcells (Fig. 4B, C) did not closely approximate that of the DNA measurements performed on the G,/G, cells of these lines (Fig. 2B, C, Table 1). Instead, the values of the CVs for both the MCa-11 and the HL-60 T-cell DNA peaks were a n order of magnitude higher than the CVs of the G,/G, DNA peaks (Table 1).Our observations were similar when we compared the M and G, DNA peaks of these lines (Figs. 2B, C and 3B, C, Table 1). Taken together, these results seem to demonstrate a new paradox: how can the MCa-11 and HL-60 mitotic cells be more DNA diverse than are their corresponding G,/Gl and G, cell populations? We believe that there are two possible, and not mutually exclusive, explanations for this apparent paradox. The first is that, by using the standard interphase DNA distribution, one may seriously underestimate the heterogeneity of the G,/G, and G2 DNA peaks for chromosomally diverse cell lines and tumors (16). Since there are no specific morphologic or staining criteria by which Go/ Gl-, S-, and G,-phase cells can be distinguished from one another, a n interphase DNA content distribution offers the potential for a significant, but undetectable, overlap among these cell populations. In a n effort to overcome this problem, DNA precursors such as BrDU and [3H]thymidine have been used for selective labeling of the S phase compartment. Such studies have shown for many tumors and cell lines, including MCa11 cells grown in vivo and in vitro, that a small proportion (-5%) of cells in the S-phase compartment do not incorporate DNA precursors even under favorable growth conditions (1,2). It is possible that some of these non-labeled cells are actually G,/G, or G, cells with non-modal DNA contents that cannot be detected in the standard DNA distribution. We do not think these non-labeled cells in the Sphase compartment are entirely responsible for the discrepancies between the M-cell and interphase-cell DNA distributions for the MCa-11 line. Even when these unlabeled cells are calculated into the CVs of the GO/G,and G, DNA distributions, these CVs increase by less than 1% (Dooley and Allison, unpublished data). Thus, even when such unlabeled cells in the S-phase compartment are accounted for, the CVs of the mitotic MCa-11 cells are still much higher than the CVs of the interphase MCa-11 cells. The increased dispersion of the MCa-11 and HL-60 M and T DNA measurements, relative to their corresponding G,/G, and G, DNA measurements (Figs. 2-4, Table l ) , may also be a reflection of varying transit times through the different phases of the cell cycle for individual cells in heteroploid cell populations. Timelapse photographic studies performed on heteroploid lines have shown that the time of mitosis is generally lengthened, and is much more variable, for cells with

105

higher chromosome numbers and karyotypic diversity (9,15,17).We have preliminary evidence that the MCa11 cells at extremes of non-modal DNA content and chromosome number have the apparent greatest difficulty in completing mitosis, with prolonged mitotic times, i.e., such cells are “metaphase delayed,” not “metaphase arrested’ (Dooley and Allison, unpublished results). We consider it possible that the nonmodal cells with prolonged mitotic times are selected for shorter times in interphase, so that their total cell cycle times lie within the range of those for cells of the modal “main stemline”, thus preventing the non-modal cells from being rapidly lost from the total cell population. This process of shortening the interphase time of the cells with non-modal DNA content relative to the mitotic time also leads to further under-representation of these cells in the G,/G, and G, peaks of standard interphase DNA distributions. The recent finding in other cell systems that certain G,/G,- and G,-phase metabolic functions can be performed in the mitotic phase of the cell cycle is consistent with this hypothesis

(5,731. In summary, we have found that the widths of the G,/G, and G2 DNA peaks, as measured in interphase DNA distributions, provide a low estimate of the degree of genetic diversity (as measured by variation in mitotic cell DNA content and/or chromosome number) actually present in two heteroploid cell lines. The DNA distribution of mitotic cells more closely parallels the chromosome number distribution of these two lines and it may therefore be a more accurate gauge of the range of genetic diversity actually present within such heteroploid cell populations. We found no evidence in either the MCa-11 or the HL-60 lines that large proportions of cells with non-modal DNA contents were arrested in metaphase or were selectively dying in mitosis. The CVs of the DNA values of the M and T cells from both of these lines showed reasonable agreement, although the CVs of the T cells were slightly lower than the CVs of the M cells (Figs. 3B, C, 4B, C, Table 1). This implies that most cells survived mitosis regardless of DNA content. Presumably, these telophase cells reenter interphase and begin another replication cycle. Our results suggest that the majority of heteroploid tumor cells, regardless of their variable DNA content and chromosome number, survive and replicate. This observation would seem to cast doubt on the tenet of stem cell theory that only cells with the modal chromosome number, and with modal DNA content, in heteroploid cell lines and tumors are responsible for the perpetual growth of these cells. Similarly, the wide spectrum of karyotypic diversity seen in some solid tumors, and the cell lines derived from such tumors, may be a reflection of the functional genetic variation within these tumors. Standard cytometric determination of tumor ploidy levels, the widths of the interphase DNA peaks, and the proportions of S-phase cells may fail in estimating the extent of this genetic diversity.

106

DOOLEY ET AL.

This genetic diversity, and its rate of formation, may possibly be very important if almost every cell has “stem cell” potential as our results suggest.

LITERATURE CITED 1. Allison DC, Anderson S, Ridolpho PF, Meyne J , Robertson J :

Alternations in the DNA metabolism of MCa-11 mouse mammary tumor cells grown in uiuo and in uitro. Cancer Res 46:39513957, 1986. 2. Allison DC, Bose KK, Anderson S, Curley S, Robertson J: Slowing of cell cycle traverse for cells in exponential monolayer culture placed into plateau-fed and starved medium. Cancer Res 49: 1456-1464, 1989. 3. Allison DC, Lawrence GN, Ridolpho PF, OGrady BJ, Rasch RW, Rasch EM: Increased accuracy and speed of absorption cytometric DNA measurements by automatic corrections for nuclear darkness. Cytometry 5:217-222, 1984. 4. Allison DC: Refinement in absorption-cytometric measurements of cellular DNA content. In: Advances in Microscopy. Alan R. Liss, New York, 1985, pp 167-185. 5, Bissen ST, Weisblat DA: The durations and compositions of cell cycles in embryos of the leech. Helobdella triserialis. Development 105:105-118, 1989. 6. Dooley WC, Roberts JR, Allison DC: Acid Giemsa for rapid identification of mitotic cells, J Histochem Cytochem 37:1553-1556, 1989. 7. El-Alfy M, Leblond CD: Long duration of mitosis and consequences for the cell cycle concept, as seen in the isthmal cells of the mouse pyloric antrum. I. Identification of early and late steps of mitosis. Cell Tissue Kinet 20:205-213, 1987. 8. El-Alfy M, Leblond CP: Long duration of mitosis and consequences for the cell cycle concept as seen in the isthmal cells of the mouse pyloric antrum. I1 Duration of mitotic phases and cycle

stages, and their relation to one another. Cell Tissue Kinet 20: 215-226, 1987. 9. Fardon JC, Prince JE: A comparison of the ratios of metaphase to prophase in normal and neoplastic tissues. Cancer Res 12:793795, 1952. 10. Hsu TC, Moorhead PS: Chromosome anomalies in human neoplasms with special reference to the mechanisms of polyploidization and aneuploidization in the HeLa strain. Ann NY Acad Sci 63:1083-1094, 1956. 11. Kraemer PM, Peterson DF, VanDilla MA: DNA constancy in heteroploidy and the stem line theory of tumors. Science 174:713717, 1971. 12. Levan A: Chromosomes in cancer tissue. Ann NY Acad Sci 63: 774-794, 1956. 13. Makino S: Further evidence favoring the concept of the stem cell in ascites tumors of rats. Ann NY Acad Sci 635318-830, 1956. 14. Makino S: Neoplasia. In. Human Chromosomes. Igaku Shoin, Tokyo, 1987, pp 429-467. 15. Rondanelli EG, Magliulo E, Pilla G, Falchi F, Barigazzi GM: Chronology of the mitotic cycle of acute leukemia cells. Acta Haematol 42:76-85, 1969. 16. Shackney SE, Burholt DR, Pollice AA, Smith CA, Pugh RP, Hartsock RJ: Discrepancies between flow cytometric and cytogenetic studies in the detection of aneuploidy in human solid tumors. Cytometry 11:94-104, 1990. 17. Sisken J E , Bonner SV, Grasch SD, Powell DE, Donaldson ES: Alterations in metaphase durations in cells derived from human tumors. Cell Tissue Kinet 18:137-146, 1985. 18. Stich HF, Steele HD: DNA Content of tumor cells. Quantitative genetic analysis of tumor progression. Cancer Met Rev 4:173194, 1962. 19. Stich HF: The DNA content of tumor cells, 11. Alterations during the formation of hepatomas in rats. J Natl Cancer Inst 24:12831296, 1959.