Cell Tissue Kinet. (198 1) 14, 489-499.

C L O N A L V A R I A T I O N I N P R O L I F E R A T I O N R A T E O F C U L T U R E S O F GPK C E L L S

P. A . RILEY A N D M. H O L A

Department of Biochemical Pathology, University College School of Medicine, London

(Received 3 March 1980; revision accepted 18 January 1981)

A B S T R A C T

Pedigrees of twenty-six clones of a line of keratocytes derived from guinea-pig ear epidermis (GPK cells) were analysed from time-lapse film. The mean interdivision time (IDT) for the culture was 1143 k 215 (SD) min. The mean generation rates (mean reciprocal interdivision times) of clones varied over a range of 3.93-10.2 x 10-4/min and the standard deviation of the clonal mean generation rates was 16.8% of the average value. Transient intraclonal variations in IDT due to mitoses in a plane perpendicular to the substratum were observed. The data were also analysed on the basis of cell location in sixteen equal zones (quadrats) of the filmed area. The mean generation rate of quadrats was 8.73 x 10-4/min (SD = 4.9%). The spatial distribution showed some clustering of cells. The mean local density of the clones (2.25 f 0.62 cel l~/ lO-~ cm2) was significantly higher than the quadrat density (1.76 k 0-8 ~el l s / lO-~ cm2). There was no significant correlation between clonal density and mean generation rates, whereas for quadrats a significant negative correlation was found (P = 2.7%). The results support the proposition that cell lineage is the major determinant of the proliferation rate of subconfluent cultures.

Previous studies of the pedigrees of diploid fibroblasts in monolayer culture by Absher, Absher & Barnes (1974) and Absher & Absher (1976) have emphasized the importance of the clonal lineage in ordaining relative rates of proliferation, and similar clonal differences in the generation times of a line of epithelial cells have also been observed (Riley & Hola, 1980a). In this paper we have attempted to examine further the relative contributions of the clonal lineage and of the local environmental conditions in the determination of the interdivision times of subconfluent monolayer cultures of guinea-pig keratocytes.

M A T E R I A L S A N D M E T H O D S

The cells examined in this genealogical study were keratocytes derived from the ear epidermis of black guinea-pigs and are referred to as GPK cells (Riley, 1980). The stock cells were

Correspondence: Dr P. A. Riley, Department of Biochemical Pathology, Faculty of Clinical Sciences, University College School of Medicine, University Street, London WC 1 655.

0008-8730/8 1/1000-0489%02.00 0 198 1 Blackwell Scientific Publications 489

490 P. A . Riley and M. Hola routinely subcultured by trypsinization in a 1 :400 split ratio and grown at 37OC in polystyrene flasks with 2.5 mM Hepes-buffered minimum essential medium (Eagle), supplemented by 10% foetal bovine serum and antibiotics (penicillin 1000 u% and strepto- mycin 1 mg%). The cultures were checked at intervals for mycoplasma contamination by autoradiography and broth culture. Cells at passage 24 (i-e., after approximately 200 cell doublings) were inoculated at a mean initial density of 2 x lo3 cells/cmz in a polystyrene culture flask. Filming was carried out as previously described (Riley & Hola, 1980a) with a lapse interval of 1 min between frames using a x6 phase contrast objective on an inverted microscope located in a 37OC room. The field area observed was 5.35 x lop3 cmz and contained twenty-nine cells at the beginning of the film (equivalent to about twice the average initial propagdle). The negative of the film was analysed by projection onto a screen divided into sixteen rectangles of equal area (3.34 x cm2); the position of the cells in relation to the grid, and the time and location of mitoses was recorded and entered on pedigree charts. The total length of the film was 8700 frames of which 7000 frames were used in the analysis. The times to the first division were recorded but not used in the analysis. The total cells per field increased exponentially during the period analysed. The final 1700 frames of the film were unsuitable for analysis due to difficulty in accurate identification of cells. Analysis of clustering was carried out for the first 4000 frames using the method of Snedecor (1956). Statistical evaluations were performed by computer using standard programs. The chi-square test of significance of deviation from the Gaussian density distribution of the generation rates was based on the procedure previously described (Riley & Hola, 1980a).

R E S U L T S

Distribution of generation times

The overall distribution of generation times is summarized as a frequency distribution histogram (Fig. 1). The mean value of the 469 samples of interdivision times was 1143 min with a standard deviation of 215 min. The skewed distribution could be fitted to a normal probability density using a class interval of 50 min (chi-square = 19.84, degrees of freedom = 24) by omitting the upper four values (Lee, those from clones 1 1 and 17).

Intraclonal variation in intermitotic interval The distribution of intermitotic intervals for each of the twenty-nine clones stemming from

progenitor cells present in the initial field, is illustrated in Fig. 2 and the variability of generation times within each clone is shown by the frequency distribution histograms. Some of the variation is accounted for by the tendency for the initial pair of generation times to be longer than subsequent values recorded for the clone, and this also partly accounts for the skewed distribution of generation times. However, even when these initial values are excluded from the frequency distribution analysis (except in the cases of clones 1 1 and 17, where only the initial values were available), a non-Gaussian distribution of generation times was found for half the clones. The deviation from normality could be corrected by taking the reciprocal of the generation times, i.e. the corresponding generation rates (Kubitschek, 1962) in compiling the frequency distribution and these data were employed in statistical comparisons (Tables 1 and 2). Table 1 shows the mean generation rates of twenty-six clones. Clones 4, 7

Clonal variation in proliferation rate 49 1

Generation time (min)

FIG. 1. Frequency distribution histogram of the interdivision times determined in the present analysis and plotted with a class interval of 50 min. The abcissa gives the generation times in minutes and the mean is indicated by the arrow.

and 28 are excluded as the progenitor cells failed to proliferate. The standard deviations range from 9 to 20% of the mean with an average of 12.5% of the mean generation rate (MGR) of the clone. Those cases exhibiting a transient variation in interdivision time (IDT) in the pedigree were examined. Close study of the mitoses preceding lengthened IDTs revealed that there is a correlation between the IDT and the length of time that the affected cell takes to flatten out on the substratum after mitosis. Generally, this resulted from cell divisions that took place in the vertical plane leaving one division product adherent to the lower cell but with no attachment to the solid substratum. This often resulted a substantial delay before the upper cell gained attachment to the culture surface and spread out. Although the delay in flattening was not equivalent to the length of the increase in IDT, in 80% of the cases examined it was the upper cells of vertical divisions that manifested the prolongation in generation times. By contrast, horizontal mitoses gave rise to division products which flattened out at the same rate and exhibited closely similar IDTs.

Inter-clonal variation The mean generation rates of individual clones ranged between 3.93 and 10-20 x 10-4/min

with a mean value of 8.44 x 10-41min. The standard deviation of the means was 1.42, i.e. 16.8% of the mean value ( n = 26).

492 P. A . Riley and M. Hola

3 um

22 1 nm

23 n m

11 I 24 n n m n n D J I ~

13 26 m n n

Generation time (min)

Fro. 2. Frequency distribution histograms of generation times of clones 1-29 (excluding clones 4, 7 and 28 which failed to proliferate in the zone of observation). The clonal means are indicated by arrows and the products of the first mitosis identified by cross-hatching.

Variation within quadrats The frequency distribution of generation times for each of the sixteen quadrats analysed

are shown in Fig. 3 and the corresponding MGRs are listed in Table 2. The relative uniformity of the distribution of mitoses in the field is indicated by the number

of observations for each quadrat (29.31 f 7.6). The standard deviations about the individual means for the locations averaged 17.7% with a range between 13 and 31% of the MGR for the quadrat.

Variation between quadrats

A relatively small variation of MGR was found for the sixteen quadrats, ranging from 7.94 to 9.43 x 10-4/min with an average value of 8 - 7 3 x 10-4/min. The standard deviation of the means was 0-43, i.e. 4.9% of the average value of the quadrat MGRs (n = 16).

Clonal density Analysis of the spatial distribution of the cells showed that clustering took place in the

Clonal variation in proliferation rate TABLE 1. Generation rates determined for clones

Mean generation SD of SD as Clone rate x lO-‘/min mean % mean No. observations

493

1 2 3 S 6 8 9

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 29

9.41 10.20 8.38 8.14 9.99 9.91 9.62 9.08 3.93 7.30 9.73 8.74 9.32 8.96 5.11 9.72 7.91 8.85 7.4s 7.61 8.66 8.12 7.72 8.43 8.91 8.46

1 .os 1.06 1.51 1.66 1.1s 1.39 I .07 0.79 -

0.89 1.12 0.86 0.79 0.81

1.50 1.22 1.27 1.21 0.7 I 1.37 1.41 1.13 0.86 1.72 1.41

-

11 10 18 20 12 14 11 9

12 11 10 9 9

1s 15 14 16 9

16 17 1s 10 19 17

-

-

23 20 29 18 25 20 27 26

2 10 26 17 23 12 2

23 18 20 14 10 18 29 15 23

9 10

culture (Fig. 4). To examine the possible influence of variations in clonal density on the clonal generation rates, approximate clonal densities were estimated by defining the clonal centre as the central point of an area circumscribed by the smallest circle including the cells belonging to that clone. Very little net movement of the clonal centres was observed. For comparison with the data from the quadrat analysis the clonal density was operationally defined as the cell numbers within a circle having an area equivalent to a quadrat (3.34 x cm2) centred on the clonal centre. The mean clonal densities were estimated by counting the total number of cells in each of the clonal areas at 1000, 2000, 3000 and 4000 min. The results are shown in Table 3. The overall mean clonal density was 2.25 & 0.61 cells per cm2 (n = 26).

Quadrat density Mean quadrat densities were estimated from the total cells per quadrat at 1000,2000,3000

and 4000 min (Table 4). The overall mean quadrat density was 1.76 k 0.8 cells per cm2 ( n = 16). This value is significantly lower than the mean clonal density ( P = 3.4%) and reflects the inhomogeneity of spatial distribution in the culture. The total number of cells occupying the field under observation was greater than the sum of the recorded clonal sizes. This was due to two factors; first, the emigration of clonal progeny and secondly the immigration of cells from the surrounding zone. Although the degree of mixing was relatively

494 P. A . Riley and M. Hola

1 IC

" n

IVA

IVB

1 IVC

1 IVD

r r -

FIG. 3. Frequency distribution histograms of generation times analysed by quadrats identified by row and column matrix. The means are indicated by arrows.

small (21% migrant cells at 4000 min) the possibility that the loss of clonal progeny could introduce a bias in the incomplete pedigree has to be borne in mind and this difficulty might be eliminated by the use of circumscribed culture areas (Carter, 1967) of appropriate size. In all cases where loss of cells from the clone being analysed occurred, this took place in an interval that was shorter than the average IDT for that generation and we have, therefore, neglected the data relating to lost cells. The influence of cellular exchanges at the margins of the filmed area on the density considerations is not considered significant since both clonal and quadrat cell densities were calculated on a similar basis.

D I S C U S S I O N

The results establish that, while there is relatively little variation within clones, there is considerable variation between the generation rates of different clones. This variation is more than three times greater than the corresponding variation between quadrats and is not due to

Clonal variation in proliferation rate TABLE 2. Generation rates of quadrats

495

Mean generation SD Quadrat rate x lO-'/min SD (% of mean) No. observations

1 2 3 4 5 6 7 8 9

10 11 12 13 14 15 16

8.85 8.96 9.00 8.73 7.94 9.15 9.43 8.56 8.06 8.93 9.13 8.18 8.29 8-92 8.98 8.64

1.79 1.54 1.54 2.12 1.68 1.25 1.35 1.11 1.36 1.40 1 *43 1 *68 1.32 1.38 0.85 2.66

20 17 17 24 21 14 14 13 17 16 16 21 16 15 10 31

31 30 31 24 21 31 37 27 41 42 31 16 23 30 18 36

I I I I 1000 2000 3000 4000

t h i n )

FIG. 4. Clustering analysis of cell distribution in the observed field. The ordinate gives the value of the parameter (P) calculated from the Poisson distribution. The interrupted line indicates the value corresponding to a random distribution of entities in the field; homogeneity is indicated by values less than unity and clustering by values exceeding unity. Analyses at intervals of 1000 min indicate an initial nearly random distribution with subsequent clustering.

differences in clonal density. Regression analyses of mean generation rate and mean density show that there is no significant relationship between these parameters for clones (correlation coefficient = -0.1; P = 33%), whereas for quadrats a significant negative correlation was found (correlation coefficient = -0.5; P = 2.7%). These results are entirely consistent with the suggestion that the major factor determining the generation rate of the cells in subconfluent culture is the cell lineage and that local environmental influences are of secondary importance.

Transient variation in the IDT were observed as shown in the pedigree (Fig. 5). Similar

496 P. A . Riley and M. Hola TABLE 3. Mean clonal cell densities

Clone Mean cell density (cell~/lO-~ cm2)

1 2.01 2 2.46 3 1.56 4 1.86 5 2.37 6 1 *86 7 8 2.01 9 2.16

10 2.98 11 2.37 12 1.41 13 1.31 14 2.38 15 1.71 16 1.49

-

- 17 18 1.64 19 2.75 20 2.83 21 3.35 22 2.68 23 2-68 24 2.23 25 1.71 26 1.93 27 3.94 28 29 1-56

-

TABLE 4. Mean quadrat cell densities

Mean cell density Quadrat (~ells/ lO-~ cm2)

A 1 1.79 I1 3.87 111 2.90 IV 2.16

B I 1.12 I1 1.71 111 1.71 IV 1.04 CI 1.86

I1 1.41 111 1.41 IV 1.64

D I 1.34 I1 1.78 I11 0.30 IV 2.16

Clonal variation in proliferation rate

1040

1350

1270 -

'-1 llao , 1000 ;; 1040

-1 L 1540

1010

1280

1290 L

FIG. 5. Pedigree chart of clone 3 showing transient variation of generation times by pairs of division products. The vertical distances are arbitrary; horizontal distances are proportional to the interdivision times indicated in minutes above the line. The cross indicates an unresolved intermitotic interval.

_ _ _ _ 1880

_ _ _ _ 1710

___- 1700

_ _ _ _ 1580

I , 1110 i""l 1120 1 1200 , 1080 ;

1200 1090

L

FIG. 6. Pedigree of interdivision times of clone 5 showing a dichotomy between the slow-growing (upper) subclone and the more rapidly proliferating (lower) subclonal progeny. The interrupted lines indicate cells that were followed for 3200 min and that the next cell division had not occurred by the end of the period analysed.

497

498 P. A . Riley and M . Hola observations were made by Absher et al. (1974) and Absher & Absher (1976) on WI-38 cells, and they have developed a model to explain the phenomenon based on metabolic exchanges between adjacent cells (Absher & Absher, 1978). It has also been proposed that the duration of the cell cycle is determined by a random transition by which a cell passes from an ‘A state’ in the G, phase to a ‘B phase’ that includes the rest of the cycle (Smith & Martin, 1973). The tendency, in a cumulative plot, for differences in the IDTs of sibling cells to be exponentially distributed has been interpreted as support for the transition probability hypothesis (Minor & Smith, 1974; Shields & Smith, 1977; Shields et al., 1978; Brooks, Bennett & Smith, 1980) although this analysis has been disputed (Castor, 1980). In our cultures transient variation in generation times was found to be associated with the mode of cell division. Cells exhibiting increased generation times were often observed to emanate from vertical- or ‘piggy-back’ mitoses. A delay in reattachment of the uppermost cell to the substratum and a lengthening of the intermitotic interval of that cell in comparison to its sibling was a feature of this type of cytokinesis (Riley & Hola, 1980b).

The causes of the variation in generation rate between clones are not known but the lengthening of the generation time of a subline of clone 5 as an apparently inherited characteristic (Fig. 6) suggests that somatic mutation, or a process with similar features, could give rise to variant clones of the type represented by clones 11 and 17, and minor differences in clonal generation rates might arise in the same manner. Evidence of sister cells giving rise to different lineages has previously been documented (Shields, 1977). Since, slowly proliferating clones have fewer progeny it is clear that selective factors will operate to the detriment of clones with long generation times. In our sample, cells of clones or sub-clones of distinctly retarded generation rate constituted about 7% of the inoculum and about 2% of the cells at the end of the growth period analysed. These cells would, therefore, comprise a smaller proportion of the inoculum in a subsequent subcultivation unless they were able to proliferate at a later confluent stage when the faster growing clones had been arrested. Because of the technical difficulties involved we have not examined confluent or near-confluent cultures and cannot exclude the possibility that slowly proliferating clones manifest a diminished density dependent inhibition of growth, although it would seem to us more likely that the 3-4-fold increase in the proportion of cells with long generation times is brought about by the production of further, slowly growing variants in other clones.

We conclude that clonal genealogies demonstrate that, in GPK cultures, cell lineage is the main determinant of proliferation rate and that, in layer culture, transient variations in the IDT may arise from spatially asymmetrical cytokinesis. Permanent changes (lengthening) in the IDT of subclones have been observed and such changes may reflect a process akin to somatic mutation which may form the basis of interclonal variation.

A C K N O W L E D G M E N T S

This work was supported by the National Foundation for Cancer Research.

R E F E R E N C E S

ABSHER, P.M., ABSHER, R.G. & BARNES, W.D. (1974) Genealogies of clones of diploid fibroblasts. Cinemicrophotographic observations of cell division patterns in relation to population aging. Exp. Cell Res. 88, 95.

Clonal variation in proliferation rate 499 ABSHER, P.M. & ABSHER, R.G. (1976) Clonal variations and aging of diploid fibroblasts. Cinematographic

ABSHER, R.G. & ABSHER, P.M. (1978) Mathematical models and computer simulations of proliferation of human

BROOKS, R.F., BENNETT, D.C. & SMITH, J.A. (1980) Mammalian cell cycles need two random transitions. Cell,

CARTER, S.B. (1967) A method of confining single cells to study individual cell reactions and clone formation.

CASTOR, L.N. (1980) A G, rate model accounts for cell-cycle kinetics attributed to ‘transition probability’. Nature,

KUBITSCHEK, H.E. (1962) Normal distribution of cell generation rate. Exp. Cell Res. 26,439. MINOR, P.D. & SMITH, J.A. (1974) Explanation of degree of correlation of sibling generation times in animal cells.

RILEY, P.A. (1980) A theory of cellular senescence based on Darwinian principles in the light of Linnaeus. In: The

RILEY, P.A. & HOLA, M. (1980a) Clonal differences in generation times of GPK epithelial cells in monolayer

RILEY, P.A. & HOLA, M. (1980b) Variation in intermitotic interval between sister cells related to orientation at

SHIELDS, R. (1977) Transition probability and the origin of variation in the cell cycle. Nature, 267, 704. SHIELDS, R. & SMITH, J.A. (1977) Cells regulate their proliferation through alterations in transition probability. J .

SHIELDS, R., BROOKS, R.F., RIDDLE, P.N., CAPELLARO, D.F. & DELIA, D. (1978) Cell size, cell cycle and

SMITH, J.A. &MARTIN, L. (1973) Do cells cycle? Proc. Nat. Acad. Sci., U.S.A. 70, 1263. SNEDECOR, G.W. (1956) Statistical methods (5th edn), Iowa State College Press, Ames.

studies of cell pedigrees. Exp. Cell Res. 103, 247.

diploid fibroblast clones. J. theoret. Biol. 72,627.

19,493.

Exp. Cell Res. 48, 189.

287, 857.

Nature, 248,24 1.

Skin of Vertebrates (Ed. by R. I. C. Spearman and P. A. Riley), p. 11 1. Academic Press, London.

culture. Exp. Cell Biol. 48, 3 10.

mitosis. Eur. J . Cell Biol. 22, 3 15.

Cell Physiol. 91, 345.

transition probability in mouse fibroblasts. Cell, IS, 469.