Proc. Natl. Acad. Sci. USAVol. 78, No. 5, pp. 3053-3057, May 1981Cell Biology

Anchorage-independent growth of normal human fibroblasts(methylcellulose/hydrocortisone/high serum concentration)

DONNA M. PEEHL AND ERIC J. STANBRIDGE*Department of Microbiology, College of Medicine, University of California, Irvine, California 92717

Communicated by Donald A. Glaser, February 9, 1981

ABSTRACT Normal human fibroblasts, considered to be en-tirely anchorage dependent for proliferation, have been grown inmethylcellulose medium. The most important factor required forgrowth in suspension appears to be the use of high levels of serumand hydrocortisone. Newborn foreskin or fetal lung fibroblastsform colonies as large as 0.5 mm in diameter after 3 wk, with acolony-forming efficiency as high as 70%. Mouse 3T3 cells that donot form colonies in standard assays for anchorage-independentgrowth also grow under these conditions. Colony formation resultsafter inoculation of as few as 100 cells per 60-mm dish, and meta-phase cells have been visualized with a fluorescent DNA stain,showing that colony formation is due to division rather than ag-gregation. Fibroblasts recovered from suspension and grown asmonolayers retain a diploid karyotype and normal shape, do notform tumors upon injection into nude mice, and become senescent.Thus, the trait ofanchorage-independent growth in vitro is clearlypossessed by normal human fibroblasts and can be expressed un-der the proper conditions.

The transition of a normal cefl to a transformed one is accom-panied by a host ofphenotypic changes. One such change is theability of the transformed cell to grow unattached to a solid sub-stratum, a property that has been termed anchorage-indepen-dent growth. This phenomenon was first described by Mac-Pherson and Montagnier (1), who used growth in soft agar asa method of selecting transformed cells from polyoma-infectedbaby hamster kidney cells. This correlation was further refinedby Shin and colleagues (2, 3), who showed a strong relationshipbetween anchorage independence and the tumorigenic prop-erties of cells.

However, results from other studies dispute the general va-lidity of growth in suspension as a property ofonly tumorigeniccells. Several investigators (4-7) have shown that certain trans-formed nontumorigenic cell lines exhibit anchorage-indepen-dent growth, whereas others have reported that a number oftumorigenic cell lines retain their anchorage dependence (8).The failure to demonstrate anchorage-independent growth

of cells may be due to inadequate culture conditions. The prob-lem of creating an optimal nutrient environment for growth insuspension ofdifferent cell types has been studied only to a lim-ited degree (9-11) and is undoubtedly as important a factor asit is for growth in monolayer.

Although it remains debatable whether or not growth insemisolid medium will prove to be a good correlate of tumor-igenicity, normal human fibroblasts are considered to be en-tirely anchorage dependent for growth in vitro. In this reportwe show that normal human fibroblasts and other supposedlyanchorage-dependent cells are capable of anchorage-indepen-dent growth under appropriate conditions.

MATERIALS AND METHODSCells. The cells used in this study are listed in Table 1. They

include several strains of normal diploid human fibroblasts,National Institutes of Health 3T3-Swiss albino mouse cells,HeLa cells, and somatic cell hybrids derived from fusions be-tween human fibroblasts and HeLa cells. The foreskin fibro-blasts were isolated in this laboratory as described (15). All hu-man fibroblasts were maintained in Dulbecco's modified Eagle'smedium (DME medium) supplemented with 10% (vol/vol) fe-tal bovine serum or in MCDB 104 medium supplemented with2% (vol/vol) dialyzed fetal bovine serum. This latter formula-tion has been shown to be superior to other medium formula-tions for the clonal growth of fibroblasts on solid substrata (16).The other lines listed were maintained in DME medium sup-plemented with 10% (vol/vol) calf serum. All cells were reg-ularly tested for mycoplasma contamination by cultural methodsand by 4', 6-diamidino-2-phenylindole (DAPI) assay (17) andwere always negative. Prior to the anchorage assays, all cellswere grown as monolayers in flasks (Falcon).

Media. Several media formulations were tested in the an-chorage assays. MCDB 104 medium was prepared from stocksolutions in this laboratory as described (16), with the exceptionthat silicon, molybdenum, vanadium, nickel, tin, and linoleicacid were omitted. DME medium from GIBCO was preparedwith 6.6 g ofHepes buffer per liter and adjusted to pH 7.4 beforethe addition of sodium bicarbonate (1.2 g/liter). Eagle's mini-mal essential medium (MEM/alpha modified; KC Biological,Lenexa, KS) was supplemented with nonessential amino acidsand vitamins. Several lots ofcommercial fetal bovine serum andcalf serum from Flow Laboratories (McLean, VA), Irvine Sci-entific (Irvine, CA), and Reheis (Scottsdale, AZ) were used.Dialyzed serum was prepared as described (18). Hydrocortisone(Sigma) was solubilized in absolute ethanol; the final concen-tration of ethanol in the growth medium was 0.1% and was notdeleterious to the cells.

Methylcellulose. Methylcellulose (2.6 g; Matheson, Cole-man, and Bell or Dow) was dispersed in 30 ml of hot (approx-imately 900C) twice-distilled H20, followed by the addition of70 ml of cold (40C) H20. The solution was stirred overnight at40C, and then autoclaved for 40 min. The solution was storedat least 24 hr at 40C before it was mixed 1:1 with doubly sup-plemented medium. Five or 10 ml of this 1.3% methylcellulosemedium was then pipetted into uncoated 60-mm bacteriologicalplates (Plasta Medic or Falcon) or tissue culture dishes (LuxScientific). In some experiments, dishes were overlaid with 5or 10 ml of 0.9% Difco agar, Noble agar, or agarose in DMEmedium supplemented with hydrocortisone and fetal bovineserum.

Abbreviations: DME medium, Dulbecco's modified Eagle's medium;DAPI, 4',6-diamidino-2-phenylindole.* To whom reprint requests should be addressed.

3053

The publication costs ofthis article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertise-ment" in accordance with 18 U. S. C. §1734 solely to indicate this fact.

3054 Cell Biology: Peehl and Stanbridge

Table 1. Cells used in anchorage independent assays

Cell Characterization Ref.

HF1031HF0814 Initiated inHF0814 > Human foreskin fibroblast author'sHF1010 laboratoryHF0813 JGM2291 Human fetal lung fibroblast 123T3-Swiss albino Mouse embryonic pericyte 13HeLa D98OR Human cervical adenocarcinoma 12ESH5 Nontumorigenic fibroblast 14

x D98OR hybridESH5T Tumorigenic segregant ofESH5 11

Anchorage Assay. Cells growing in monolayer were detachedwith 0.05% trypsin (crystalline, Sigma) or with 0.025% trypsin/0.002% EDTA in phosphate-buffered saline (pH 7.4). The cellswere suspended in medium supplemented with serum, and themixture was centrifuged to pellet the cells. The cells were re-

suspended in DME medium supplemented with 10% fetal bo-vine serum, and 0.2 ml containing 102-106 cells was inoculatedinto methylcellulose medium in plates. Suspension was accom-

plished by vigorous mixing. The plates were incubated in hum-idified 95% air/5% CO2 at 37°C for up to 21 days. At the endof the assay period, colony number and size were determinedwith the use of a calibrated microscope eyepiece.

Visualization of Cell Division. At various times, colonies inmethylcellulose medium were removed, stained with DAPI(17), and mounted on glass slides under a coverslip. The DNA-binding dye DAPI readily allows the identification of cells inmetaphase. The colonies were examined with the use ofa LeitzDialux microscope equipped with Ploem fluorescent illumination.

Cells Isolated from Methylcellulose. Colonies were isolatedfrom methylcellulose medium with a Pasteur pipette and grownas monolayers. These cells were characterized with regard tomorphological characteristics, longevity, and karyotype. Tu-morigenicity was determined by subcutaneous injection intoathymic nude mice (19).

RESULTSGrowth of Normal Human Fibroblasts in Methylcellulose.

Fetal lung fibroblast cell strain GM2291 and newborn foreskinfibroblast cell strain HF1031 were inoculated into methylcel-lulose medium in uncoated bacteriological dishes. The abilityof different medium formulations to support colony formationwhen 100 cells were inoculated per dish is listed in Table 2. Thecell strain GM2291 formed colonies in all the media tested, witha colony-forming efficiency as high as 49% in DME mediumsupplemented with hydrocortisone (10 ug/ml) and 20% fetalbovine serum. Foreskin fibroblasts also formed colonies when100 cells were plated, but only in three of the six media. How-ever, when 1000 cells were inoculated per dish, colonies formedin all media (data not shown). DME medium supplementedwith hydrocortisone and 20% fetal bovine serum supportedmaximal growth of both cell strains.The results from this experiment indicated that a reduction

in the level of serum or hydrocortisone led to a decrease in col-ony-forming efficiency and colony diameter ofboth cell strains.To further examine this point, the anchorage-independentgrowth of GM2291 cells in DME medium supplemented withvarious levels of hydrocortisone was evaluated. Although sizewas relatively unaffected, the colony-forming efficiency in-creased significantly with an increase in hydrocortisone con-

centration (Table 3).The medium conditions then were extended to include

Table 2. Growth of normal human fibroblasts in methylcellulosemedium

GM2291 HF1031

Colony size Colony size

Aver- Aver-CFE, age, Range, CFE, age, Range,

Medium % mm mm % mm mm

MEM/10% CS 15 0.13 0.10-0.18 0 - -DME/10% CS 11 0.13 0.10-0.18 0 -

DME/10% FBS 9 0.19 0.13-0.25 0 - -DME/10%FBS/10 ,g ofHC per ml 26 0.16 0.08-0.25 4 0.15 0.10-0.18

DME/20% FBS 40 0.24 0.10-0.43 12 0.11 0.10-0.15DME/20%FBS/10 ,tg ofHC per ml 49 0.20 0.13-0.35 77 0.17 0.10-0.25

Cells of each strain were inoculated at 100 cells per dish into meth-ylcellulose medium in 60-mm uncoated bacteriological dishes. Colony-forming efficiency (CFE) and size were determined after 21 days of in-cubation. For each medium, all colonies in one dish were counted andsized. MEM, minimal essential medium; CS, calf serum; FBS, fetalbovine serum; HC, hydrocortisone.

MCDB 104 medium, a formulation developed for optimal clonalgrowth of fibroblasts on solid substrata. MCDB 104 supportedminimal colony formation, even when used in conjunction withhydrocortisone and fetal bovine serum (Table 4). This findingwas unexpected in view of the superiority of MCDB 104 me-dium to support optimal colony formation in monolayer culture(16). It was thought that the high level of calcium present inMCDB 104 medium (1 mM) might be the major factor inhibitinggrowth in suspension, but a reduction of the level of calciumto 0.5 mM did not improve growth. It is possible that the con-centration of one or more nutrients in MCDB 104 medium istoo low to support growth in suspension, because other typesof cells that are capable of growth in both monolayer and sus-pension require higher concentrations of nutrients in the latersituation (R. G. Ham, personal communication; ref. 20).

Influence of Substratum upon Colony Formation in Meth-ylcellulose. In order to exclude the possibility that colonies wereforming on the surface of the dish rather than in suspension,bacteriological dishes were overlaid with DME medium sup-plemented with 10% fetal bovine serum and containing 0.9%Difco agar, Noble agar, or agarose. Colony formation ofGM2291 cells was almost completely eliminated in dishescoated with Difco or Noble agar (Table 5). However, coloniesdid form in agarose-coated dishes, albeit at a lower frequencythan that seen in the uncoated dishes. In separate experiments,similar results were seen with strains HF1010 and HF0813 (datanot shown).

Table 3. Effect of hydrocortisone on growth of human fibroblastsin methylcellulose medium

Colony size

HC, ,ug/ml CFE, % Average, mm Range, mm

0 7.6 0.14 0.08-0.230.1 36.0 0.15 0.05-0.251.0 50.4 0.13 0.05-0.28

10.0 55.2 0.15 0.05-0.28GM2291 cells were inoculated at 250 cells per 60-mm bacteriological

dish into methylcellulose medium containing DME medium supple-mented with 20% fetal bovine serum (FBS) and the indicated levels ofhydrocortisone (HC). After 21 days of incubation, all colonies in eachdish were sized and counted. CFE, colony-forming efficiency.

Proc. Natl. Acad. Sci USA 78 (1981)

Proc. Natl. Acad. Sci. USA 78 (1981) 3055

Table 4. Effect of basal medium and supplements uponanchorage-independent growth

Supplements Colony size

HC, FBS, CFE, Average, Range,Medium llg/ml % % mm mm

DME 10 20 22.5 0.14 0.05-0.3MCDB 104 2* 1.7 0.09 0.05-0.13MCDB 104

(5 x 104M Ca2+) 2* 0.3 0.09 0.08-0.1MCDB 104 10 20 13.4 0.12 0.08-0.25MCDB 104

(5 x 10-4M Ca2+) 10 20 6.9 0.13 0.08-0.2

GM2291 cells were inoculated at 1000 cells per 60-mm uncoated bac-teriological dish into methylcellulose media as indicated. All coloniesper dish were sized and counted on day 21. CFE, colony-forming effi-ciency; HC, hydrocortisone; FBS, fetal calf serum.* Dialyzed fetal bovine serum.

Similar results were seen when tissue culture instead of bac-teriological dishes were used (Table 6). Colony formation in tis-sue culture dishes was slightly decreased. In particular, almostno colony formation occurred in tissue culture dishes whenmethylcellulose with DME medium and only 10% fetal bovineserum was used. Colony formation in bacteriological or tissueculture dishes overlaid with agarose occurred only if DMEmedium supplemented with 20% fetal bovine serum was usedand was greatly increased by the addition of hydrocortisone.

Progressive Growth of Colonies in Methylcellulose. GM2291cells were inoculated at 1000 cells per uncoated bacteriologicaldish containing methycellulose in DME medium supplementedwith hydrocortisone (10 ,g/ml) and 20% fetal calf serum.Dishes were selected on various days after inoculation, and thediameter of all colonies was measured. Fig. 1 shows the pro-gressive increase in colony size with time. On day 1, only singlecells were visible (not illustrated). Small colonies of 0.05-mmdiameter were noted on day 5. Colony-forming efficiency re-mained essentially the same between days 9 and 21, whereasthe colony size steadily increased.

These results indicated that colony formation and growth wasdue to active division rather than to aggregation ofsmall coloniesor cells. To further verify this, representative colonies wereisolated and stained with DAPI to visualize metaphase cells. Arepresentative colony is shown in Fig. 2a, and Fig. 2b. showsa colony stained with DAPI. At least one metaphase spread wasseen in most colonies, indicating mitotic activity. Owing to thelimitation of this technique, only metaphases at the peripheryof a colony could be identified. A crude estimate of the numberof cells in a colony was obtained by gently squashing the indi-vidual colonies under a coverslip and counting the dispersed

Table 6. Effect of medium on anchorage-independent growth inagarose-coated sishes

DME medium Colony-forming efficiency, %supplements Uncoated Agarose underlay

HC, FBS, Bacteri- Tissue Bacteri- Tissue,.g/ml % ological culture ological culture- 10 1.1 0.3 0 010 10 3.1 1.4 0 0- 20 5.3 4.3 0.7 0.310 20 14.6 10.1 4.4 5.1

GM2291 cells were inoculated at 1000 cells per dish into medium inbacteriological- or tissue culture-grade dishes. The dishes were eitheruncoated or overlaid with 10 ml of agarose medium. Methylcelluloseor agarose medium was prepared with DME medium supplemented asindicated. Colonies were counted on day 14.

nuclei stained with DAPI. In most instances it was estimatedthat a minimum of eight population doublings had occurred inthe 21-day period.

100

60

20

100

60

20u,

0

U4

04-4

U)~o

~0z

100

60

Day 5

Day 9

Day 13

20

Day 1780

40

Table 5. Effect of agar and agarose on anchorage-independentgrowth

Colony size

CFE, Average, Range,Surface % mm mm

Uncoated 22.5 0.14 0.05-0.3Difco agar 0 - -Noble agar 0.7 0.09 0.05-0.13Agarose 9.5 0.10 0.05-0.2

GM2291 cells were inoculated (1000 cells per dish) into bacteriolog-ical dishes overlaid with 5 ml of agar or agarose medium. Methylcel-lulose medium was prepared with DME medium supplemented withhydrocortisone (10 ,ug/ml) and 20% fetal bovine serum. Colonies werecounted and sized on day 21. CFE, colony-forming efficiency.

Day 21

60

20 km15 0.2 0.25 0.Colony diameter, mm

.3 0.35 0.4

FIG. 1. Progressive growth of colonies of GM2291 cells in meth-ylcellulose medium. One thousand cells were inoculated into each 60-mm bacteriological dish containing 1.3% methylcellulose in DME me-dium supplemented with hydrocortisone (10 ,ug/ml) and 20% fetal bo-vine serum. Dishes were selected on the days indicated, and all colonieson each dish were counted and sized.

Cell Biology: Peehl and Stanbridge

3056 Cell Biology: Peehl and Stanbridge

FIG. 2. Colonies isolated from methylcellulose medium. (a) A 21-day colony of GM2291 cells grown in 1.3% methylcellulose in DMEmedium supplemented with hydrocortisone (10 gg/ml) and 20% fetalbovine serum. (Phase contrast; X1200) (b) Colony of GM2291 cellsgrown as in a and stained with DAPI. Arrows indicate cells in mitosis.(x460.) (Inset) Higher magnification of one of the mitotic cells.(x2000.)

Characteristics of Cells Recovered from Methylcellulose.HF0814 cells were grown in methylcellulose with DME me-

dium, hydrocortisone (10 ,ug/ml), and 20% fetal bovine serum.On day 21, all colonies from one dish were recovered and grownin monolayer in DME medium supplemented with 10% fetalbovine serum. When sufficient cells were obtained, 3.5 X 106cells were injected subcutaneously into each of two sites on

nude mice. No tumors have formed after more than 4 mo.

GM2291 cells were grown in methylcellulose as describedfor the HF cells. On day 21, an individual colony was isolatedand grown in monolayer. The cells were typically fibroblasticin morphological characteristics, and chromosome analysis re-

vealed a normal diploid karyotype (data not shown). The cellpopulation became senescent after approximately 43 populationdoublings.

Growth of Other Anchorage-Dependent Cells in Methyl-cellulose. It has been reported that 3T3-Swiss albino cells donot form colonies in semisolid medium (3, 4). Therefore, theywere tested under conditions that were optimal for the an-

chorage-independent growth ofhuman fibroblasts in this study.The colony-forming efficiency of the 3T3 cells was low (3.4%)with an average colony size of 0.1 mm (range, 0.05-0.2 mm)(Table 7). This degree of colony formation is significantly lessthan that seen with the human fibroblasts but is more than thatreported in other systems (3, 4).

For comparative purposes, cells that had been shown to be

Table 7. Growth of other cells in methylcellulose medium

Colony sizeCell line CFE, % Average, mm Range, mm3T3 3.4 0.1 0.05-0.2D98OR 98.0 0.14 0.05-0.25ESH5 14.3 0.31 0.08-0.63ESH5T 49.6 0.29 0.08-0.75

In separate experiments, 102 to 104 cells were inoculated into 60-mmuncoated bacteriological dishes containing methylcellulose with DMEmedium supplemented with hydrocortisone (10 ,ug/ml) and 20% fetalbovine serum. Colonies were sized and counted on day 21. CFE, colony-forming efficiency.

anchorage-independent in other studies were tested underthese conditions. The results are included in Table 7. HeLa cellsformed colonies similar in size to those formed by human fi-broblasts but with an extremely high colony-forming efficiency(99%). The nontumorigenic fibroblast X HeLa hybrid ESH5and its tumorigenic segregant ESH5T formed colonies with anefficiency similar to that reported in other systems (6).

DISCUSSIONMost anchorage assays have involved the use of bacteriologicalor tissue culture dishes overlaid with agar. Under these con-ditions, lack of growth of normal human fibroblasts has beenuniformly documented, and the cells have been presumed tobe entirely anchorage dependent for growth. The loss of thisanchorage dependence, although not always correlated withtumorigenicity, is nevertheless one of the common character-istics of transformed cells. In this regard, the use of semisolidmedium to selectively grow stem cells of human tumors hasbeen fruitful (21).The results reported in this paper indicate that simple and

seemingly minor modifications in these conventional systemscan create an environment in which cells shown to be normalby several criteria (i.e., morphology, karyotype, mortality, andnontumorigenicity) can divide and form progressively enlargingcolonies in an anchorage-independent environment. Six strainsof fibroblasts besides those reported here have formed coloniesunder the conditions described (data not shown).

The conditions favoring growth of normal human fibroblastsin methycellulose appear to be the use of medium supple-mented with hydrocortisone and a somewhat higher level ofserum than is routinely used. The use of high levels of serumand hydrocortisone is necessary for optimal colony formation.When dishes were overlaid with agarose prior to the additionof methylcellulose, colonies formed only when DME mediumsupplemented with 20% serum was used. The most importantpoint is that colonies did form in methylcellulose when therewas an agarose underlay. This finding excludes the possibilitythat colony formation occurred at the surface of the dish ratherthan in suspension. The complete lack ofcolony formation whenan agar underlay was used is most probably due to toxic factorsin the agar that are known to inhibit the growth of cells (22).

It has been reported that normal rat fibroblasts can be in-duced to grow in an anchorage-independent fashion. Todaroand de Larco (23) isolated a growth factor from medium con-ditioned by sarcoma virus-transformed mouse cells. This factorstimulates normal rat fibroblasts to divide in semisolid mediumand produces a morphological change when added to the samecells grown in monolayer. Both the anchorage-independentgrowth and the morphological change from that of a normal fi-broblastic appearance to one of a transformed, disorganizedtype are reversible upon removal of the growth factor.

Proc. Natl. Acad. Sci. USA 78 (1981)

Proc. Natl. Acad. Sci. USA 78 (1981) 3057

At the present time, the reasons why normal human fibro-blasts are capable of anchorage independent growth in our sys-tem are not completely known. The conditions supporting op-timal growth in our experiments [DME medium supplementedwith hydrocortisone (10 ,g/ml) and 20% serum] have not toour knowledge been tested before. In most studies, tissue cul-ture grade dishes are overlaid with agar or agarose followed byaddition of the suspension medium containing the cells underinvestigation. Also the media used are usually supplementedwith considerably less than 20% serum (6, 24, 25). The lack ofanchorage-independent growth offibroblasts noted under theseconditions also was obtained in our studies. It is possible, there-fore, that the normal fibroblast cells are intrinsically capable ofsuch anchorage-independent growth and that in previous in-vestigations the conditions tested were inhibitory to these cells.Alternatively, the conditions to which we have exposed the cellsin our assay may induce the cells to synthesize the growth factorsnecessary for independent growth. These possibilities are underinvestigation. The conditions described in this study may them-selves not be optimal for sustained anchorage-independentgrowth. For example, although the HeLa cells formed colonieswith high efficiency, the average size of the colonies was no

greater than that observed for human fibroblasts.It is clear that normal human diploid fibroblasts must now

join the list of normal cells capable of anchorage-independentgrowth. This list includes hematopoietic cells (26) and endothe-lial cells (27). Furthermore, it is clear that anchorage-indepen-dent growth, as an intrinsic property, is not confined to trans-formed or tumorigenic cells and, thus, poses a warning toinvestigators utilizing this assay in clinical studies of drug sen-

sitivities of putative human tumor cells growing in semisolidmedium (21).Our observation that single normal cells will initiate DNA

synthesis and divide when suspended in methylcellulose is con-trary to the hypothesis ofFolkman and Moscona (28), who statethat the shape of a cell is critical for DNA synthesis by normalcells and that when normal cells assume a spherical shape, as

in suspension, they are unable to proceed through DNA syn-thesis. In our studies the individual fibroblast cells assume a

spherical shape in methylcellulose and, of course, proceed todivide actively. Interestingly, when an obvious colony isformed, the cells on the periphery of the colony (Fig. 2b) lineup in a fibroblastoid manner.

Finally, the finding that normal human diploid fibroblasts are

capable of anchorage-independent growth and yet retain theirnormal phenotypic characteristics will perhaps stimulate rein-vestigation of the conditions necessary for growth of these cellsin liquid suspension. Accomplishment of this would greatly en-

hance the usefulness of human fibroblasts for human viral vac-

cine production and also may lead to new concepts ofthe controlof growth of normal cells in culture.

These studies were supported by U.S. Public Health Service GrantCA19401. E.J.S. is recipient of U.S. Public Health Service ResearchCareer Development Award 1 K04 CA 00271. D. M.P. was supportedby a postdoctoral training grant (GM 07307) awarded by the NationalInstitutes of Health.

1. MacPherson, I. & Montagnier, L. (1964) Virology 23, 291-294.2. Freedman, V. H. & Shin, S. (1974) Cell 3, 355-359.3. Shin, S.-I., Freedman, V. H., Risser, R. & Pollack, R. (1975)

Proc. Natl. Acad. Sci. USA 72, 4435-4439.4. Stiles, C. D., Desmond, W., Chuman, L. M., Sato, G. & Saier,

M. H., Jr. (1976) Cancer Res. 36, 3300-3305.5. Gallimore, P. H., McDougall, J. K. & Chen, L. B. (1977) Cell 10,

669-678.6. Stanbridge, E. J. & Wilkinson, J. (1978) Proc. Natl. Acad. Sci.

USA 75, 1466-1469.7. Jones, P. A., Lang, W. E., Gardner, A., Nye, C. A., Fink, L. M.

& Benedict, W. F. (1976) Cancer Res. 36, 2863-2867.8. Marshall, C. J. (1979)J. Cell Sci. 39, 319-327.9. Colburn, N. H., Bruegge, W. F. V., Bates, J. R., Gray, R. H.,

Rossen, J. D., Kelsey, W. H. & Shimada, T. (1978) Cancer Res.38, 624-634.

10. Pollack, R., Risser, R., Conlon, S. & Rifkin, D. (1974) Proc. Natl.Acad. Sci. USA 71, 4792-4796.

11. Stanbridge, E. J. & Wilkinson, J. (1980) Int. J. Cancer 26, 1-8.12. Weissman, B. & Stanbridge, E. J. (1980) Cytogenet. Cell Genet.

28, 227-239.13. Todaro, G. J. & Green, H. (1963)J. Cell Biol. 17, 299-313.14. Stanbridge, E. J. (1976) Nature (London) 260, 17-20.15. Peehl, D. M. & Ham, R. G. (1980) In Vitro 16, 516-525.16. McKeehan, W. L., McKeehan, K. A., Hammond, S. L. & Ham,

R. G. (1977) In Vitro 13, 399-416.17. Russell, W. C., Newman, C. & Williamson, D. H. (1975) Nature

(London) 253, 461-462.18. McKeehan, W. L., Hamilton, W. G. & Ham, R. G. (1976) Proc.

Natl. Acad. Sci. USA 73, 2023-2027.19. Stanbridge, E. J., Boulger, L. R., Franks, C. R., Garrett, J. A.,

Reeson, D. E., Bishop, D. & Perkins, F. T. (1975) Cancer Res.35, 2203-2212.

20. Paul, D. (1978) Exp. Cell Res. 114, 434-438.21. Hamburger, A. W. & Salmon, S. E. (1977) Science 197, 461-463.22. Montagnier, L. (1971) in CIBA Foundation Symposium on

Growth Control in Cell Cultures, eds. Wolstenholme, G. E. &Knight, J. (Churchill, London), pp. 3341.

23. Todaro, G. J. & de Larco, J. E. (1978) Cancer Res. 38,41474154.

24. Kahn, P. & Shin, S.-I. (1979)J. Cell Biol. 82, 1-16.25. Kopelovich, L. & Bias, N. E. (1980) Exp. Cell Biol. 48, 207-217.26. Metcalf, D. (1977) in Recent Results in Cancer Research, ed.

Rentchnick, P. (Springer, New York), Vol. 61, pp. 1-227.27. Laug, W. E., Tokes, Z. A., Benedict, W. F. & Sorgente, N.

(1980) J. Cell Biol. 84, 281-293.28. Folkman, J. & Moscona, A. (1978) Nature (London) 273, 345-349.

Cell Biology: Peehl and Stanbridge