NATURE. VOL. 220. OCTOBER 19. 1 968

We thank Dr K. Hirschhorn and Mrs S. Villafane for blood samples from cases 1 and 4.

This work was supported by grants from the Population Council to A. de C. and from the Health Research Council of the City of New York and the National Institute of Child Health and Human Development, US Public Health Service, to 0. J. M. and W. R. B.

A. DE CAPOA D. A. MILLER 0. J. MILLER

Department of Obstetrics and Gynecology, College of Physicians and Surgeons, Columbia University, New York.

W. R. BREG Southbury Training School and Department of Pediatrics, Yale University. Received July 15,1968. ' De Capoa, A. , Breg, W. R., Eushnick, T., and Miller, 0. J., Ann. Hum.

Genet., 32,191 (1968). ' Giannelli, F., and Howlett. R. M., Cytoflenetics, 6, 186 (1966). ' Warburton, D., Miller, 0. J., Breg, W. R., de Capoa, A,, and Shaw, M. W.,

Amer. J . Hum. Genet., 19, 399 (1967). Utakoji, T., and Hsu, T. C., Cytogeneties, 4, 295 (1965). Bishop, A,, Leese, M., and Blank, C. E., J . Med. Genet.,%, 107 (1965). ' Turpin, R., and Lejeune, J., in Les Chromosomes Humains, 64 (Gauthier-

Villars, Paris, 1965).

spontaneous transformants in cultures of normal cells, or cultured from cancerous tissue, were much less susceptible than diploid cells to growth inhibitory effects and attained three to ten-fold greater population densities. Strains of hypodiploid and trisomic human fibrobIasts behaved like normal cells in their susceptibility to population density inhibition3.

Human cell strains which were susceptible to growth inhibition in pure culture usually inhibited each other in mixed culture3. (An exception was the absence of in- hibition between early passage epithelial human amnioil cells and human fibroblasts, including those deriving from amnion.) Conversely, cells which had escaped in greater or lesser degree from inhibition of growth a t high popula- tion densities were inhibited to only a minor degree when superinoculated on to fully grown and essentially static cultures of diploid cells. These experiments with human, monkey, mouse, rabbit, vole and hamster cells concern the degree of species specificity in the growth inhibitory effects of cellular interaction.

The cell lines used in this work are listed in Table 1. The methods of cultivation have been previously descrlbed4, as has the procedure of superinoculating a formed cell monolayer in replicate T-15 flasks with cells of a different type3. Cell lines were chosen for the underlying substrate (Table 1) on the basis of two criteria: relatively low maximum population density in terms of protein content or cell counts, and the ability of the stabilized culture to survive 4-10 days after superinoculation without coming off the glass.

The number of cells in the culture was determined in a Coulter particle counter. The culture medium was removed, the cells washed quickly with growth medium7 lacking serum, calcium and magnesium (MB) and thorl treated with the same medium containing versene a t 0.04 per cent and Difco trypsin a t 0.25 per cent. This was removed after 15-30 s. When the cells had separated from the glass and had largely dissociated, they were

Species Specificity in Growth Regulatory Effects of Cellular Interaction SURFACE (monolayer) cultures of human diploid cclls show a marked decrease in the rate of synthesis of DNA, RNA and net protein as the population density of the cultures increases1, but it has not yet been determined whether growth inhibition occurs because of actual cellular contact, or whether growth inhibitory compounds are released extracellularly to act only a t very short range2. Heteroploid epithelial human cells, either arising as

Table 1. CELL CULTURES USED FOR STUDY OF INTERSPECIFIC GROWTH INHIBITORY EFFEClR

Maximum* Maximum* cells rsed for under- Used for Strain protein (gg/cn12) x lo4 (per cmz) lying substratest s~~pcrinornlation Species

Human Diploid fibroblasts HG 46 E- 699

WI 26 Penny BAL Det. 510 E 196 Det. 551 W I 38

Heteroploid Spontaneous "Liver" transformants FL

RA9 Sarcoma RPMI 41s

RPMI 1922s Epithelial cancer HeLa

cells E B C 4%

African green monkey Kidney AGMK (early):[ WGM CV-1 AGMK-CP

Mouse 3T3 L S 180 Epithelial Fibroblast-likc BHK

Lens11 nlploid Heteroploid

ICidney Lung (embryonic) Lung t t

Rabbit

Hamster

Vole

* Highest values attained in three-ten experiments. t Cells highly susceptible to growth inhibition at high population density. Saturated cultures were used as substratesfor superinoculation with other

cell lines for the determination of heterospecific inhibition. t Symbols used in Figs. 1-3 to designate superinoculated cells. Open symbols ( 0, A , 0. 0) represent cells extremely susceptible to growth inhibition

at high population density. All were dlplold exceptfor 3T3. Half-Alledsymbols (9, A , P, +, v)represent spontaneously altered, heteroploid cells dcriving from normal tissues; and sohd symbols (a, A) represent cells c u l t u r o m cancers.

5 RA-a "triploid" culture of human amnion cellsK. RPMI 41 1922-cultures of human sarcomata'. C &human cancer culturev. a AGWK (early) were green monkey kfdney oells in the first thiee passages and still diploid. I I Lens (dipl01d)--early passage ep~thel~al cells isolated by Dr Arnold Shapiro. Lens-a fibroblast-like, spontaneously altered and heteroploid variant

of the diploid cell. ** Misleading low value, caused by separation of cellsfrom glassin crowded cultures. t t Cultiired by Dr Eva Hansen-Melander. The provenance Of cllltures not otherwise specified has been described'.

NATURE, VOL. 220. O C T O B E R 19. 1968

Table 2 . PARTIAL INHIBITION OF A MOUSE CANCER CELL (S 180) SUPERINOCULATED ON TO A 6 DAY CULTURE OF EARLY PASSAGE APRICAN QREEN MOXKEY KIDNEY CELLS (AGMK)

Substriltc alone (AGMK)

Superinoculated cells alone ( S 180)

Combined culture

Net growth of superinocu- Control lated cells

In mixed culture

pg protein Cells x loL pg protein Cells x 10" pg protein Cells x lo4

Protein Cells Protein Cells

Protein and cell determinations On day of On day after superinoculation

superinoculation 1 2 4 6

489 524 556 616 597 57 55 57

9 6 t 557 762 1,oon 20 93 146 166

569 773 984 116 150

t The size of the inoculum was determined by protein analysis and cell counts 24 h after planting and includes the usually negligible growth of cells in that initial 24 h. A quantitatively more serious potential error is introduced by the continuing slow growth of the cell8 in the underlying monolayer (compare with top row of table). The calculation assumes that the underlay grew to the same (minor) extent in the mixed culture as i t did in the control. I n this example, in the unlikely event that the substrate monlayer had itself been completely inhibited by the relatively small super~noculum, the percentage of superinorulum growth after 4 days would be g84-g6-489 = 60 per cent of tho contro1,instead of 41 per cent.

762-96

shaken for 5-10 s in 5 ml. of MB-versene medium (no trypsin) until they were almost totally monodisperse, when 5 ml. of complete growth medium containing 10 per cent serum was added, and the suspension counted. Samples were diluted 1 : 2 or 1 : 3 if counts were in excess of 2 x 10"ml. to minimize coincidence error. Counts in replicate flasks were similar within 5 per cent.

The degree to which superinoculated cells adhered to the monolayer, and their persistence during the following 5-10 days of incubation, was determined by using cells prelsbelled with 14C-thymidine, added to the medium a t 0.025 pCi/ml. for 4-8 days before superinoculation. The superinoculated cells usually adhered to the underlying monolayor as effectively as in a control empty flask (Table 3). In a few instances the adhesion was enhanced. In most experiments the labelled inoculated cells per- sisted in the mixed culture throughout the period of incubation. In some experiments, the I4C counts of the superinoclilated cells decreased progressively but just as rapidly in the control as in the mixed culture, probably because of partial separation of cells from the glass in the course of the incubation period. In some cases this decrease may have been caused by the incorporation of 14C-thymidine into the DNA of mycoplasma which may have then dissociated from the celllo. No difference was noted, however, between cultures infected with myco- plasma and those uninfected with respect to their ability either to inhibit or to be inhibited by other cell types.

The possibility that the superinoculated cells might have adversely affected the underlying monolayer was examined by prelabelling the underlying cells for 3-6 days with 3H-thymidine a t 0.025 pCi/ml. before superinocula- tion. In most experiments, the underlying monolayers were not affected by the growth of the superinoculated cells: the 3H-thymidine counts were sustained in the mixed culture. Occasionally, however, there was a marked decrease in tritium activity. No attempt has been made to determine whether the decrease was caused by cell turnover or mycoplasma in the underlaylo, but it was usually as rapid in the control cultures as in those which had been superinoculated.

At the end of the experiment the cells were dissolved in 114 strength Lowry's solution C and portions were used to determine cell protein by a modified Lowry procedure8. Samples (0.5 ml.) were counted in a Beckman scintillation counter, the channel windows set so that approximately 6 per cent of the 14C count spilled over into the aH channel, wh11e none of the 3H channel counts appeared in the 14C channel.

An illustrative experiment showing a partial inhibition of superinoculated mouse cancer cells (S 180) by a green monkey cell monolayer is summarized in Table 2.

In Figs. 1-3 the degree to which various cell strains grew when superinoculated on to fully sheeted and slowly growing monolayers of the same or different species has been expressed as a percentage of the growth attained by the same inoculum in pure culture. The results based on cell counts and proteins were averaged.

With the human diploid cells listed in Table 1 as sub- strates (Fig. 1) there was a reasonably good correlation between the ability of superinoculated cells to escape the inhibitory effects of an underlying monolayer and their inherent growth potential in pure culture (Table 1). This was unrelated to the animal species of the super- inoculated cells. Cells with low maximum growth potential were usually completely inhibited, whether human, rabbit or mouse. Superinocula labelled with 14C (Table 3) showed that they had adhered to the underlying mono- layer, and had persisted during the following 5-10 days' incubation. The observed inhibition of growth was there- fore caused by the failure of the adherent superinoculated cells to multiply.

In contrast, heteroploid human, mouse, hamster or vole cancer cells or spontaneously altered cells derived from normal tissues, which were all capable of attaining high population densities in pure culture, were only

Table 3. SUPERINOCULATED CELLS WHICH BAILED TO GROW IN COPBINED CULTURE* USUALLY ADHERED TO THE UNDERLYINO YONOLAYER?, AND

PERSISTED IN THE FOLLOWING 5-10 DAYS' INCUBATION

Cell in underlying substrate

Superinoculated 14C-labelled cell

Human E 699 Mouse 3T3 3,349 3,483 2,531 3,085 diploid Det 510 3,349 1,910 1,990 2,190 (Fig. 1) Det 510 Rabbit lens (epithelial, 277 370 314 307

W I 38 diploid) 3,060 2,870 3,315 3,030 WI 38 HeLa 24 23 22 11

RPMI-41 2,080 1,980 754 699

Monkey-CV (Fig. 2) Human-E 699 33 23 35 19 Mouse-L 2,766 1,221 f 2,426 6645 Rabbit lens 45 29 - 24

Mouse-3T3 (Fig. 3) Monkey-CV 142 130 120 93 -CP 25 26 - -

Rabbit lens 1,088 1,153 1,033 874

Rabbit lens Human Penny 497 546 348 408 diploid Det. 510 905 1,156 1,106 1,048

HG 46 577 460 587 512 Det. 532 1 0 9 0 785 1,120 1,090

Monkey CV 1:790 1,262 1 6 0 0 1,200 CP 1,471 716f 1:435 2858

Mouse 3T3 2,208 1,958 1,956 1,312 * Growth of su~erinoculated cells < 10 per cent that of same cells in pure

culture. f Same asin a control culture inoculated directly on to glass. $.Experiments in which failure of superinoculated cells to grow was

attributable in part to their failure to adhere to the underlying substrate a t the time of inoculation.

g Experiments in which failure of superinoculatcd Cells to grow was attributable iq part tqtheir progressive dissociationfrom the substrate during the follomng ~ncubatlon.

NATURE. VOL. 220. OCTOBER 19. 1968

50 100 160 200 250 "Growth potential" of superinoculated cell

(pg protein/cme In pure culture)

Fig. 1. The growt,h inhibitory effects of "contacted" human cell snb- strates on snperinocolat,ed cells. Fully sheeted and saturated cult.uws of human diploid cells were superinoculated with small nnmhers of a varlety of other cell types (0, human; A, mouse; 0, monkey; ,0, vole; x , hamster; 0, rabbit). Hnlf closed symbols are heteroplo~d variants of normal cells, and solid symbols are cells cult.ured from cancers (compare with Table 1). The net growth of the superinoculated cells, expressed as a percentnpr of that observed in a pure culture, has been plotted as a f~mction of their "growth potential" (maxlmum density achieved in pure culture in the conditions of these experimcnts (Table 1)). e*, H u m n ~ ~

cancer cells inhibited by human diploid strain WI 38.

slightly inhibited, as indicated in tho right hand portion of Fig. 1. (A 50 per cent growth inhibition in a com- bined culture after 5 days, when thc control had increased* for example, sixteen-fold, reflects only a slight prolonga- tion in generation time, from 1.25 to 1.66 days.) A I ~ exception was the ability of one human diploid cell strain (WI 38) to inhibit certain human cancer cells (HeLa, 41), not inhibited by other human diploid strains (Fig. 1). It is possible tlrat this is related to the higher. cell density (and protein content) achieved by t,his WI 38 monolayer (Table 1).

Intermediate results were obtained with overlays of heteroploid monkey and rabbit cells, corresponding to their intermediate capacity for independent growth.

When a mouse fibroblast monolaycr (3T3) was super- inoculated with a variety of cell lines, there was again a correlation (Wig. 2) between the growth potential of the superinoculated ccll and the degree to which it escaped from the inhibitory effect of the underlying and contact inhibited monolaycr. As with human coll substrates, the degree of inhibition was largely independent of tho species of the superinoculated cell.

Somewhat different results were obtained with three green monkey kidney strains as substrates (Fig. 3). Although some cell overlays were inhibited, again w~thout, regard to the species of origin, there was no clear inverse correlation between tho inGerent growth capacity of the superinoculated cell and its susceptibilitv to inhibition bv thk underlying monolayer. A Aouse iine (3T3) which showed marked contact inhibition in pure culture (Fig. 3) was not greatly inhibited by the monkey cell monolayer. Conversely, several heteroploid human lines with a high growth capacity were more inhibited than some diploid human cells which attained smaller population densities in pure culture. The L mouse cell line, capable of attaining high population densities (Fig. 3), was, nevertheless, .strongly inhibited by the monkey cell substrate. This was

partly causcd by the failure of this specific overlay to adhere fully to the monkey cell substrata, and its continu- ing dissociation during the following incubation (Table 3).

An unexpected result was obtained with two lines of rabbit lens as substrates, one a diploid epithelial cell which showed a high degree of contact inhibition, and the other a heteroploid fibroblastic variant capable of attaining con- siderably higher population densities (Table 1). Both substrates almost completely inhibited the growth of super- inoculated human fibroblasts, monkey kidney cells and mouso 3T3 cells; both were partially inhibitory for vole cells; and neither strain inhibited hamster cells signifi- cantly. When these two strains woro superinoculated with heteroploid human or malignant mouse cells, however, the strongly contact inhibited diploid rabbit cell was less effective than its heteroploid counterpart in imposing its own growth arrost on to the superinoculated cells (Table 4).

A variety of human diploid cell strains highly sus- ceptible to self inhibition in pure culture1 inhibit each other in mixed cultures, while heteroploid human cells which in pure culture are insensitive to growth regulation (and which in consequence attain high population den- sities) are inhibited only slightly when superinoculated on to an arrested confluent culture of human diploid cells3. I n these experiments the inhibition of a number of human, mouse, hamster, rabbit, vole and monkey cells by under: lying oell substrates was not species specific: cells of the same and different species were equally inhibited. The ability of superinoculated cells to escapo from the growth inhibitory effects of human, mouse or rabbit coll substrates was inversely related to the inherent growth potential of the superinoculated cell, regardless of its specics; and the population density of a cell type in pure culture was inversely related to its susceptibility to inhibition by substrate monolayers of other cell types (Figs. 1 and 2). This relationship suggests that cancer cells, and many spontaneous heteroploid transformants of norrnal cells, have escaped from corltact inhibition of growth because they fail to respond to the norrnal inhibitory stimulus. The possibility that such cclls may continue to emit that inhibitory signal is not excluded.

The correlation between the growth potential of a given cell and its susceptibility to inhibition by other cell types was not complete and thero were several instances

1 I I I

50 100 160 200 250 300 "Growth potential" of superinoculated oell

(pg proteinlcm' in pure culture)

Wig. 2. The growth inhibitory effects of "contacted" mouse cell sub- strates on superinoculated cells (compare lcgend to Fig. 1).

NATURE. VOL. 220, OCTOBER 19, 1968

of selective inhibitory interactions (Table 5). Several cancer cell inocula which were not significantly inhibited by most human diploid cells were markedly inhibited by cell line WI 38 (Fig. 1). Similarly, a mouse heteroploid fibroblast (L) which had a high capacity for growth, and which was not significantly inhibited by, for example, human fibroblasts, was markedly affected by monkey kidney cells (CV) (Figs. 1 and 3).

Conversely, the 3T3 mouse fibroblast, self-inhibited in pure culture and correspondingly sensitive to inhibition by human and rabbit cells, was not inhibited by monkey kidney cells (strain CV) (Fig. 3); and the latter, which were strongly inhibited by each other, mouse cells and rabbit cells, were not inhibited by human diploid cells (Fig. 1).

There was no clear correlation between the capacity of a cell strain to inhibit other superinoculated cells, and its own susceptibility to contact inhibition. The W I 38 strain, which attained higher population densities and cell protein per unit area than any of the human diploid strains tested, was, nevertheless, the most effective irr inhibiting the growth of superinoculated heteroploid cells (Fig. 1 ) Similarly, although a (largely) diploid epithelia,] rabbit lens cell was the most self-inhibited of any tested. reaching maximum population densities of only 3.5 x lo4 cells and 31 pg of cell protein/cm2 of culture surface, it) was considerably less inhibitory for superinoculated cells bhan a heteroploid fibroblastic variant, which was sig- nificantly less contact inhibited and which attained 12 x 104 cells and 102 pg of oell protcin/em2 (Table 4). The latter was the most actively inhibitory of all the cultures tested, and significantly reduced the growth rate of a number of superinoculated human, mouse and volc lines which were affected only slightly by contact with other cell substrates. At least in these two examples ~~nusually marked inhibitory activity of a cell substrate for other cell types did not depend on a high degree of self-inhibit,ion.

I t would be of interest to determine whether cancer cells, which are relatively insensitive to growth inhibition by other cell types, are still capable of inhibiting the growth of superinoculated normal cells. Experiments on

"Growth potential" of auporinoculated cell (pg protetn/cma In pure culture)

Table 4. DIrrpEEENTm GROWTH INBIBITORY EITUCTS OQ DIPbOID AXD BETEROPLOID RABBIT LENS OEL& SUBSTRATES ON SUPEELNOOULATED EUPAH

AND MOUSE CELL8 Growth relative to that in pure culture (%) when si~perinocnlated

Cell type superinoculated onto rabbit cell ellbstrnte Diplaid Hetvroploid

epithelial* Lbrobl~stict Human Spontaneous RA

transformants FL 69 3.11 Epitbelloid cancer HeZa 59 0, 6, 20

KB 80, 100 40.41 Sarcomata RPMI 41 45 14 36

RPMI 1,922 40 19: 19 Mouse L 60.75 lo. 18. 22

8 180 109 30.50 * Rabbit lens cell which attains densit of 3-5 x 10' cells and 31 pg protein

per cma cultlve surface (compare Table 1 Footnote). Spontaneously altered variant which attains density of 11.8 x 10' cells

ulture surface. and 102 pgprotein per cm c

Table 6. 8ELECTIVITY IN AOXO AND HETERO-BrROIRIO GECWTH INHIBITORY EE'FECTB Cell species superinoculated

Cell species Human Mouse in underlying Diplold Hetcroploid Monkey 3T3 L

subatrats Growth of sunertnoculated celie, relative to that tn - pure culture, per cent

Human Fibroblast O* 55 50 8 78 diploid lines (0-;6) (35-95) (25-100) (0-12) (20-80)

WI 38 411 alone

Monkey

Mouse (ST3)

Rabbit Diploid 0 64 (0-19) (40-100) (2) 60, 75

Aneuploid 0 17 1 1 181 1 ( 0 ) (W41) (632) (10-22)

* Median of four to fifteen ohservattons wlth varlety of cell lines; range in parentheses.

the interaction of virus trunt;formants with normal cclls arid with each other and the species specificity of these effects are being carried out.

In summary, human, mouse, rabbit and monkey cells sensitive to contact inhibition of growth usually inhibit, each other, regardless of species. Cancer cclls (or spon- taneous transformants of normal cells) which have escaped from contact inhibition of growth, and in consequence achieve high population densities in pure culture, are correspondingly insensitive to inhibition by other cells of the same or other species. It appears that such cells have to different extents lost their capacity to respond to the growth inhibitory signal. Exceptions to both these generalizations have occurred with specific cell combinations.

We thank Miss Mina Levy and Miss Gail Donaldson for technical assistance. This investigation was supported by a Public Health research grant, US National Institute of Allergy and Infectious Diseases, and by grants from the American Cancer Society and the US National Science Foundation.

HARRY EAGLE EL~,IOT M. LEVINE

Division of Biological Sciences, Albert Einstein College of Medicine, Bronx, New York.

The Wistar Institute, Philadelphia, Pennsylvania.

Received June 26; revised Aumst 12,1968.

Levine, B. M., Becker, Y., Boone, G . W., and Eagle, H., Proc. US Nut. Acad. Scz., 53, 350 (1965).

a Stoker, M. P., and Rubin. H., Nature,216.171(1967). a Eagle, H., and Lcvine, E. Y., Nature,213,1102 (1967). ' Eagle, H., W%9hington, C., and Friedman, 5. M., Proc. US Nut. doad. Sci..

56,166 (1066). iI6 Regan, J. D., and Smith, J. B., Scknce,l49,1516 (1965). 1. Moore, G. E., Mount, D., Tara, G., and Schwartz, N., Cancer Re8.,28,1?'35

I1 QfiR) ,-"--,. Auersberg, N., J. Nat. Cancer Znst., 32,135 (1984). ' Eagle, H., Scisncs,l80,432 (1959).

Ovama. V. I.. and Each. H.. Proc. Soc. Exu. Biol. and Med.. 91.305 (1956). Fig. 3. The growth inhibitory effect8 of "contacted" nlolrkey cell

*ohstrates on supcrinoculated cells (oon~pnre legend to Fig. 1).

- * - - , , - - ,-, lo Levine, E. M., Thomas, L., McGregor, D., Hayflick, L., and Eagle, H., Proc.

US Nut. Acad. Sm., 60, 583 (1968).