comparison of rates of phenotypic variability in surface antigen … · of appearance of...

[CANCER RESEARCH 43, 4291-4296, September 1983]

Comparison of Rates of Phenotypic Variability in Surface Antigen

Expression in Normal and Cancerous Human BreastEpithelial Cells1

Jerry A. Peterson,2 Robert L. Ceriani, Edward W. Blank, and Lydia Osvaldo

Bruce Lyon Memorial Research Laboratory, Children's Hospital Medical Center, Oakland, California 94609 [J. A. P., R. L C., E. W. B.], and Pathology Department John

Muir Hospital, Walnut Creek, California 94598 [L 0.]

ABSTRACT

A method is described for measuring the rate of phenotypicvariability in normal and neoplastia breast epithelial cells. Threegroups of normal human mammary epithelial cells were studied,two derived from reduction mammoplasties and one derivedfrom the normal breast tissue of a patient with fibroadenoma.The breast carcinoma cells were all cell lines, four (MCF-7, SKBR-3, MDA-MB-157, and T47D) derived from pleural effusions ofpatients with breast cancer, and one (BT-20) derived from a

primary breast tumor. The heterogeneity and variability in expression of a cell surface glycoprotein with apparent molecular weightof 400,000 were studied at the single-cell level with immuno-

peroxidase techniques using a specific monoclonal antibody,BLMRL-HMFG-Mc5, to a nonpenetrating glycoprotein. The rate

of appearance of quantitative variants in expression of thisspecific surface antigen (rate of phenotypic variability) was determined in clonal colonies and was found to be severalfoldhigher in all five breast carcinoma cell lines (mean, 2.23 x 10~2/

cell/generation) than in the normal breast epithelial cells (mean,0.36 x 10~2/cell/generation). In addition, a considerable quanti

tative variation in expression of this surface antigen was demonstrated among the cells of each population in both normal andneoplastic breast cells which spread over an 8- to 10-fold range.

Furthermore, the quantitative distribution among single cells wasnot random, for the cells tended to cluster around values that fita geometric series.

INTRODUCTION

Heterogeneity and variability in expression of phenotype arecommonly remarked properties of tumor cells (for review, seeRefs. 7 and 21) and could lie at the basis of their ability in thecourse of their natural history in a patient to progress to higherstages of malignancy. Often, tumors will remain dormant andlocalized year after year and then suddenly begin to rapidlyinvade surrounding tissues and metastasize to distal sites. Inaddition, a tumor may respond initially to a given hormonal oradjuvant therapy, but after a period of time tumor growth resumes that is resistant to the therapy, and a relapse ensues.

Tumor growth and progression can be viewed as evolutionaryphenomena in populations of neoplastic cells. Theoretically, themore heterogeneous the tumor is and the greater its rate ofphenotypic variability, the more rapidly it would be expected toevolve. The most evident type of phenotypic variability is quan-

1This work is supported by Grants CA 26790 and CA 20286 from NIH, AmericanCancer Society Grant PDT-99A, and Biomedicai Research Support Grant RR05467from the Department of Health, Education, and Welfare.

2To whom requests for reprints should be addressed.

Received January 10,1983; accepted May 25,1983.

titative, where quantitative variants in many enzymes and othercell products can be readily obtained by simply cloning andisolating subclonal populations (19). Normal tissues generallymaintain a well-defined and stable phenotype, although they also

exhibit heterogeneity when analyzed quantitatively (11) at thesingle-cell level. Therefore, since single-cell heterogeneity is

found in both normal tissues and tumors, the difference betweenthem may lie in their rate of phenotypic variability. We postulatethat this single-cell heterogeneity is the result of an intrinsic

variability in cellular phenotype that is ever present in all tissues.According to this hypothesis, an imporant change that occursupon neoplastic transformation is an increase in the rate ofphenotypic variability that results in an increased statistical probability for the appearance of variants, some of which may havea more aggressive phenotype. With rat hepatoma cells, we haveadopted the Luria-DelbrÃ¼ck fluctuation test (15) in order to

determine the rate at which variants in serum albumin contentappear in clonal colonies and found it to be on the order of 10~2/

cell /generation (18). Since quantitative variation could play animportant role in the progressive evolution of tumor cells, we setout to compare the RPV3 of tumor cells with that of their normal

counterparts.The human breast carcinoma system was chosen for this

study as an example of a solid tumor with a wide variability innatural history (13). The advantages of this system are thattechniques have been developed for obtaining relatively purecultures of normal breast epithelial cells (25), the cultures can bereadily identified as breast epithelial cells with specific heterolo-

gous (3) and monoclonal (2) antisera, and there are proceduresfor the cloning of single breast epithelial cells (23).

We have already described the population heterogeneity withregard to both cell surface antigens and other components ofmouse mammary tumors (12). From one tumor, it was possibleto isolate 5 different cell populations that grew in vitro and thatdiffered from each other in phenotypic expression of the severalparameters studied (12). In the human breast, we also havedescribed in flow cytofluorimetry studies a quantitative heterogeneity in cell surface antigen expression for both normal andneoplastic epithelial cells (20). Therefore, we decided to investigate the quantitative variation of one specific cell surface component that we have defined with a monoclonal antibody, Mc5,that was prepared using delipidated human milk fat globules asthe immunogen (2). The antigen identified by Mc5 is termedNPGP, is associated with secretory capabilities of epithelial cells,and is present in high levels on human milk fat globule mem-

3The abbreviationsused are: RPV, rate of phenotypic variability; Mc5 (BLMRL-HMFG-Mc5), monoclonal antibody to a nonpenetrating glycoprotein (apparent M,400,000); NPGP, nonpenetrating glycoprotein; DME, Dulbecco's modified minimal

essential medium; FBS, fetal bovine serum.

SEPTEMBER 1983 4291

on March 13, 2020. © 1983 American Association for Cancer Research.cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

J. A. Peterson et al.

branes and on the cell surface of breast epithelial cells, but noton a number of nonbreast epithelial or fibroblastoid cell lines (2).It is a large mucin-like glycoprotein with an apparent molecular

weight of 400,000 (2).In this paper, we analyzed at the single-cell level, using im

munoperoxidase techniques, the variability of NPGP content inclonal colonies of normal and neoplastic breast epithelial cells.Further and most important, we establish a general methodologyfor comparing rates of phenotypic variability in normal and neoplastic cells. In this report, we demonstrate that neoplastic breastcell lines have a severalfold higher RPV than do normal breastcells. Our results clearly separate these 2 groups of cells andsuggest a new parameter that could be of prognostic significancein breast cancer.

MATERIALS AND METHODS

Normal Human Mammary Epithelial Cells. Normal human mammaryepithelial cells were prepared by procedures described previously (25)with slight modifications. Normal tissues were obtained from 2 reductionmammoplasties (1NB and 4NB) and one mastectomy from a patient withfibroadenoma (3NB). In the latter case, only normal tissue from the glandwas used. Skin and grossly fat areas were removed, and the remainingtissue was cut into 1- to 2-mm pieces. These pieces were digested

overnight in DME plus 5% FBS, insulin (5 jig/ml), collagenase (200 units/ml; Sigma Chemical Co., St. Louis, Mo.), and hyaluronidase (100 units/ml; Sigma) at 37Â°while gently rotating. The digested material was then

filtered through a Nitex filter (Tetko, Inc., Elmsford, N. Y.) with lOO-^m

pores. Retrieved from the filter were epithelial clumps, termed organoids(25). The organoids were aliquoted and stored frozen in the presence of10% dimethyl sulfoxide. When required, organoids were thawed andplated in a 60-mm tissue culture dish. The culture medium for normal

human mammary epithelial cells consisted of 40% conditioned media[10% from BT-20 cells, 10% from human mammary fibroblast cells, and20% from colon carcinoma (HT-29) cells], and 60% DME containing 10%

FBS, penicillin (100 units/ml), streptomycin (100 iÂ¿g/m\),and Fungizone(0.25 Mg/ml). After 7 to 10 days in culture, single-cell suspensions of

breast epithelial cells were obtained by partial trypsinization of cells thathad migrated out from the organoids. For the preparation of clonalcolonies, using procedures described previously (23), the trypsinizedcells were seeded sparsely (300 to 600 cells/sq cm) on coverslipscontaining feeder layers of UV-irradiated normal human mammary fibro-

blasts.Breast Carcinoma Cell Lines. The human mammary carcinoma cell

lines MCF-7 (24), MDA-MB-157 (26), SKBR-3 (9), BT-20 (14), and T47D(1) were maintained in either Waymouth's medium or DME supplemented

with 10% FBS and antibiotics. Clonal colonies were grown using thesame conditioned medium as with the normal cells, but only in the caseof BT-20 and SKBR-3 was the fibroblast feeder layer used.

In some cases, to unequivocally demonstrate that single cells gaverise to phenotypically mixed colonies, the cells were seeded on smallcoverslips (4x11 mm), and then those coverslips with only single cellswere transferred to new Retri dishes and allowed to develop into colonies.

Immunoperoxidase Detection of NPGP. After a culture period of 3to 8 days, the cells on the coverslips were rinsed 3 times with phosphate-buffered saline (0.14 M NaCI-0.09 M NaKPO4 buffer, pH 7.4), fixed withacetone for 2 min, and stored at -20Â° until stained. The colonies were

stained for NPGP by standard immunoperoxidase techniques (5), usingas a primary antibody the mouse monoclonal Mc5 (2). The primaryantibody was then detected with a biotinylated horse anti-mouse IgG(Vector Laboratories, Inc., Buriingame, Calif.) followed by the avidin-

peroxidase system (Vector Laboratories). The rabbit antiserum againstkeratins was generously provided to us by Dr. Tung-Tien Sun, Depart

ment of Dermatology, Johns Hopkins University School of Medicine,Baltimore, Md.

The determination of the relative concentration of NPGP per cell was

carried out on the immunoperoxidase-stained cells using a microscope

spectrum analyzer (Farrand Optical Co., Valhalla, N. Y.) attached to amicroscope. For this quantitation, the monoclonal Mc5 was used inexcess at a dilution of 1:200. The relative NPGP content is expressedas absorbance units calculated from measurements of percentage trans-

mittance with 100% transmittance set on the bare coverslip.

Absorbance = log 100/T

where T equals the percentage of transmittance. We have shownpreviously for serum albumin determinations in rat hepatoma cells stainedby immunoperoxidase techniques that the absorbance values per cellsobtained are proportional to albumin content measured by other means.4

The readings with the microscope spectrum analyzer were taken in acircular area 3 Mm in diameter in the cytoplasm of the flattened breastcells just adjacent to the nucleus. Measurements were made only oncells that were spread on the surface of the coverslip to a similar extent.The values obtained were, therefore, those of concentration (absorbanceper unit area) rather than those of total NPGP per cell.

Determination of RPV. The RPV was determined as describedpreviously (18). For this, the breast cell colonies, stained for NPGP, werescored by visual inspection and divided into 2 groups, those colonies inwhich each cell appeared to be stained to a similar extent (colonies withall negative cells were included in this group) and those colonies thatwere mixed in which either some cells stained more or less intenselythan the other cells in the colony or some cells were negative and therest were positive. Mixed colonies were only those in which cells withsimilar morphology were clearly different in their staining properties (Fig.1, D and Â£).The RPV was equal to -In P0 In 2/(/V2 - A/,), where P0 was

the proportion of colonies in which each cell stained to a similar extent(including also negative colonies), N2 was the average number of cellsper colony, and A/1 was the number of cells initiating the colony, thusequal to 1.

RESULTS

Our objective was to assess quantitatively the variability inexpression of a surface antigen (NPGP) of normal human mammary epithelial cells and compare it with that of cell lines derivedfrom breast carcinomas. For this purpose, a methodology wasdeveloped to measure the RPV in clonal colonies.

NPGP was detected in single cells using immunoperoxidasetechniques (Fig. 1). Primary cultures of normal human mammaryepithelial cells from 3 different patients were initiated with epithelial organoids prepared from the breast tissues by collagenaseand hyaluronidase digestion (25). Four of the breast carcinomacell lines were derived from metastatic pleural effusions of breastcancer patients, and one (BT-20) was derived from a solid tumor.

All of the normal human mammary epithelial cells, as well as thecell lines, were confirmed to be of breast epithelial origin usingspecific heterologous anti-human mammary epithelial antisera

(3). In order to analyze the clonal variability in expression ofNPGP in the normal and tumor breast epithelial cells, single-cell

suspensions were seeded sparsely onto coverslips and allowedto develop into colonies. In every case except for the T47D cellline, 3 types of colonies developed with respect to the NPGPantigen detected by the monoclonal Mc5: (a) colonies with allnegative cells (Fig. 1, B and C); (b) colonies with all positive cellsin which each cell stained to a similar extent (Fig. 1^); and (c)mixed colonies either in which some cells stained more intenselythan others or in which some cells were negative and some werepositive (Fig. 1, D and Â£).For the cell line T47D, no negativecells were found (Table 1). Overall, for both the normal and tumor

4J. A. Peterson, W. L. Chaovapong, and A. A. Dehghan, unpublished observa

tions.

4292 CANCER RESEARCH VOL. 43



Increased Phenotypic Variability in Breast Cancer

Table 1

Comparison of RPV in expression of a cell surface antigen (NPGP) in donai colonies of normal humanmammary epithelial cells and breast carcinoma cell lines

NPGP was detected in single cells with immunoperoxidase techniques.

No. ofcells/colonyCell

sourceNormal

breast1NB3NB4NBMeanBreast

carcinomacelllinesMCF-7T47DSKBR-3MDA-MB-157BT-20MeanMean24.99.210.814.813.312.77.85.1Range3-1053-493-423-473-635-333-223-11%

of positivecells87348388100989987No.

of coloniesscoredTotal891006198100995478No.

mixed61242221028620RPV*(X1Q-2)0.420.440.230.361.280.61.962.325.032.23

" See "Materials and Methods1 for calculation of RPV." Mixed colonies are those that contain quantitative variants in NPGP content (see "Materials and Methods").

cells, there was considerable quantitative variation from cell tocell (see below). In the case of the cell line MCF-7, we demon

strated unequivocally that single cells gave rise to mixed coloniesby visually identifying small coverslips with only single cells thatthen were allowed to form colonies.

In other studies with rat hepatoma cells (18), we described amethod for determining the rate (RPV) at which quantitativevariants arose in clonal colonies. The calculation of RPV is basedon the method that Luria and DelbrÃ¼ckdevised for determiningmutation rate in bacteria (15). In the case of the breast cells, thevariants are defined as cells that have a visually discernibledifference in NPGP content. The colonies with these quantitativevariants are referred to as mixed colonies. These quantitativevariants were usually segregated into clusters within the mixedcolony (Fig. 1, D and Â£),and the clusters ranged in size from asingle cell to one-half of the colony. There were differences in

growth rate among the different normal and tumor cell cultures,but usually by 5 to 8 days colonies of appropriate size wereobtained [average colony size for each group of cells, normaland tumor, was between 5 and 24 cells (Table 1)]. Within eachof the 3 groups of normal human mammary epithelial cells, aswell as within all the cell lines, there was considerable variationin colony size (Table 1). In all cases, a few single cells anddoublets were present that continued to express the NPGPsurface antigen, but these were not included in the calculationof RPV.

The cells that do not stain for NPGP can still be considered tobe of epithelial origin. Many of them have a clearly epithelioidmorphology. Also, in the case of 4NB where 83% of the cellsstained for NPGP, 100% of cells in colonies stained for keratin.

The RPV for shifts in expression of NPGP was calculated asdescribed in "Materials and Methods." In quantitative studies

described below using the microscope spectrum analyzer, wehave established that, for an quantitative difference to be visuallydiscernible, it must be at least 2-fold. Therefore, the RPV that iscalculated describes the rate at which a quantitative shift of morethan 2-fold in NPGP content occurs per cell per generation. Thenegative cells may still contain NPGP, but at a level belowdetectability limits of the current immunoperoxidase method. Asshown in Table 1, the mean RPV for normal breast epithelialcells from the 2 reduction mammoplasties (1NB and 4NB) and

from the normal breast tissue of a patient with fibroadenoma(3NB) was 0.36 x 10"2/cell/generation. In contrast, the meanRPV for the 5 breast carcinoma cell lines was 2.23 x 10~2/cell/

generation. Moreover, each cell line taken separately was significantly (p < 0.02, by f test) above the mean of the 3 groups ofnormal cells (Table 1).

As mentioned above, we have established unequivocally thata single cell gives rise to a mixed colony. However, when scoringseveral hundred colonies, it was not feasible to verify for eachindividual colony that it actually arose from a single cell. If themixed colonies were not the result of a change in NPGP expression in one of the daughter cells of a cell division, an alternativehypothesis could be that each cell population is made up ofsubpopulations that have different levels of expression of thesebreast surface antigens and that the mixed colonies are thosethat arose for 2 or more phenotypically different cells. This isunlikely, since the manner by which the variants arose in thecolonies was consistent with their arising spontaneously in arandom fashion as determined by the fluctuation test of Luriaand DelbrÃ¼ck(15). In other words, the number of variants percolony had a high coefficient of variation, while the proportion ofvariants (e.g., negative cells or cells with a very high NPGPcontent) in the overall population remained relatively constantwith a lower coefficient of variation. As long as the averagenumber of cells giving rise to a colony is small enough so thatsome of the colonies do not have variants (P0 > 0), then theRPV can be determined as described in "Materials and Methods."

A change in A/!, from 1 to 2, would only slightly change the RPVand would have an insignificant effect on the difference betweennormal and neoplastic breast cells, since A/2 is considerablylarger. Moreover, for the breast carcinoma cell line, MCF-7, we

visually established that 80 to 90% of the cells seeded weresingle isolated cells at 4 hr after seeding. By immunoperoxidasestaining, we found that 10% of these cells were negative forNPGP. If we assume that each cell had an equal probability ofgiving rise to a colony, we can determine the frequency expectedfor mixed colonies of positive and negative MCF-7 cells by

multiplying the proportion of negative cells in the population (0.1)by the maximum proportion of colonies that arose from morethan a single cell (0.2). This calculation predicts that the maximumpercentage of colonies containing both positive and negative

SEPTEMBER 1983 4293




cells would be 2% if they arose only from clumps of cells.Experimentally, 16 of the 22 mixed colonies presented in Table1 contained negative cells, representing 16% of the total numberof colonies scored.

Due to their growth requirements, the normal breast epithelialcells were always cloned in conditioned medium and onto UV-

inactivated fibroblast feeder layers. On the other hand, 3 of the5 cell lines were cloned without feeder layers. It is doubtful thatthe dramatic difference in RPV between normal and tumor cellswas due to a feeder layer effect, since both BT-20 and SKBR-3cells would clone only with feeder layers and had RPVs 10- and5-fold higher than did normal cells, respectively. Furthermore,MCF-7 cells had similarly high RPVs, whether they were clonedin Waymouth's medium plus 10% FBS, Waymouth's medium

plus 10% FBS plus conditioned medium, or DME plus 10% FBSplus conditioned medium plus feeder layer. RPVs in the 3 different conditions were, respectively, 2.6 x 10~2, 1.29 x 10~2, and2.74 x 10"2/cell/generation.

Quantitative Variability in Surface Antigen Expression. Using a microscope spectrum analyzer attached to a microscope,we have estimated the relative NPGP content in single cells. Thevalues for NPGP content are expressed as units of absorbanceper unit area of the cell (see "Materials and Methods"). The

relative mean NPGP contents in the normal breast cells are0.156, 0.313, and 0.206 for 1NB, 3NB, and 4NB, respectively.For the breast carcinoma cell lines, they are 0.249,0.332,0.072,0.326, and 0.135 for MDA-MB-157, SKBR-3, MCF-7, T47D, andBT-20, respectively. The coefficient of variation for the 3 groupsof breast cells is 36%, and for the breast carcinoma cell lines itis 52%. This confirms our previous results that there is considerable quantitative variation in expression of breast epithelialsurface antigens among different breast cell lines (22). In thepresent study, we have further analyzed the quantitative variationamong single cells within each population. As illustrated in Chart1 for the normal cells, 1NB, and the cell line, SKBR-3, the NPGPcontent per cell varies within each cell population over an 8- to10-fold range.

We have shown previously that clonal variation in synthesesof many proteins in hepatomas and cell lines is not random butrather tends to be among values that fit a geometric series inwhich consecutive values differ by V2 (17, 19). More recently,we have shown that the single-cell variation in albumin content

among both hepatoma cells and normal hepatocytes is over awide range (10-fold) but that the distribution is not a Gaussian

curve but rather consists of regularly spaced subpopulations thatcan be fit to a V2 geometric series.4 In the case of NPGP content

in both the normal and neoplastic breast epithelial cells (Chart1), the distributions are also not Gaussian (p < 0.01 by x2 test

for goodness of fit) but rather appear to consist also of regularlyspaced peaks along the logarithmic scale on which the NPGPcontent is plotted (Chart 1). In order to determine if there is anyregular periodicity in the distribution of NPGP contents as suggested by the presence of regularly spaced peaks (Chart 1), wehave devised a rigorous objective method for analyzing theperiodicity. This method involves the use of grids with vertical,parallel openings (windows) that are spaced at regular intervals.One such grid (shaded areas) is illustrated in Chart 1-4where thewidth of the openings is equal to that of one bar of the histogram,and the distance between the centers of each window is a factorof 1.41 (\/2) on the logarithmic scale of the abscissa. Sincephenotypic variation along a \/2 geometric series has been

/

0.025 0.05 O.I 0.2 0.4

NPGP ContentI.3

j/XAAAA/

0.05 O.I 0.2 0.4 0.8

NPGP ContentChart 1. Single-cell distribution of NPGP content in normal and neoplastic breast

epithelial cells. The cultured cells were stained with the monoclonal antibody, Mc5,using immunoperoxidase techniques, and NPGP content was determined in singlecells with a microscope spectrum analyzer. The wave patterns on the right weregenerated by the grid tests for periodicities of factors of 1.3,1.41 (or â€¢JZ),and 1.6in the distribution of NGPG content (see "Materials and Methods"). The shaded

areajn A represents a portion of the grid used to test for a periodicity of a factorof V2. The interval between vertical openings in this grid is a factor of vi on thelogarithmic scale of the abscissa A, normal breast epithelial cells (1NB); 8, breastcarcinoma cells (SKBR-3).

observed so frequently and for so many cell products (19), wechose to analyze a factor of \/2 periodicity as well as slightlysmaller (1.3) and slightly larger (1.6) factors. By moving the gridhorizontally and counting the number of circles of the histogram(i.e., number of cells) that appear in the windows at differentpositions and then plotting these sums versus consecutive positions, we can generate a wave pattern (right-hand portions of

Charts 1A and 16). One cycle of the wave pattern representsthe movement of the grid from the starting point until the latterreappears in a window. If there is a periodicity in the distributioncorresponding to that of the test grid, a regular wave pattern willbe obtained. The sharper the peaks are in the distribution, thegreater will be the amplitude of the wave pattern. If the periodicityof the test grid does not fit that of the distribution, then a skewedor perturbed wave pattern will be obtained. In every case inwhich we have applied this test to the distribution of NPGPcontent in breast epithelial cells (normal and tumor), a regularwave pattern was obtained with the 1.41 grid. A regular wavepattern was also obtained with a 1.3 grid. The regular wavepattern with the greatest amplitude was generated with thenormal breast epithelial cells using the 1.41 grid (Chart 1>*).The1.6 grid gave a perturbed wave pattern in all cases (Chart 1).Similar results were obtained when the single-cell distribution in

albumin content in normal hepatocytes was compared to that ofhepatoma cells.4

DISCUSSION

The heterogeneity of mammary epithelial cells cultured frommammary tumors, both in experimental animals and in humans,has been reported previously (12, 13, 20, 22). However, in nostudy has the heterogeneity of the corresponding normal mam-




Increased Phenotypic Variability in Breast Cancer

mary epithelial cells been quantitatively compared. In the presentstudy, we demonstrate a cell-to-cell heterogeneity in expression

of a specific breast epithelial cell surface antigen on both normaland neoplastia human breast epithelial cells, which is shown tobe both qualitative and quantitative in nature. The quantitativeantigenic distribution follows a pattern that we have previouslytermed geometric phenotypic variability (19) where the distribution of antigen content per cell is discontinuous, with cellsclustering around values that fit a geometric series in whichconsecutive values differ by >/2. This pattern of variability hasbeen observed for many cell components of other tissues in bothtumors and cell lines (19).

The most important conclusion of the present report is thatnormal human mammary epithelial cells in short-term culture can

be distinguished from established breast carcinoma cell lines bytheir RPV. Five different breast carcinoma cell lines all have RPVsseveralfold higher than normal. Most probably, the increasedRPV is a result of neoplastic transformation, although from thepresent studies we cannot eliminate the possibility that it is aresult of the cell lines being all long-term cultures. Studies arecurrently under way on short-term cultures of breast carcinoma

cells to clarify this matter.The hypothesis that acquired or increased genetic lability can

play a role in tumor progression has been discussed previously(16). An increased mutation rate has been reported to be associated with increased metastatic potential (4) and double viraltransformation (10) in cell lines, whereas no change was notedin chemically transformed human fibroblasts (6). Since in theselatter studies mutation rate was equated with acquisition of drugresistance and these rates were on the order of 10~6, these

authors may not be evaluating the same type of instability as inour studies, which involved quantitative shifts of 2-fold or moreand RPVs on the order of 10~2. Further studies are necessary

to distinguish among genetic, epigenetic, and regulatory factorsthat may be responsible for these phenotypic instabilities.

The acquisition of higher RPVs could be related to an event incarcinogenesis that is referred to as "initiation" (8). Tumor initia

tion does not involve the immediate appearance of a tumor butclearly results in an increase in the probability of a tumor laterdeveloping in the affected tissue. The specific surface antigen,NPGP, that is analyzed in this study is most probably not involvedin the neoplastic process itself, but rather its increased rate ofvariability reflects general instability in control of phenotypicexpression of all or most cellular genes.

The methodology described in this paper for determining RPVis simple and reproducible. At the basis of the method lies theability to prepare clonal colonies of breast epithelial cells asdescribed previously (23). To this must be added the use of areagent (e.g., monoclonal antibody) and a procedure (e.g., im-munoperoxidase techniques) for detection, at the single-cell level,

of a specific cell component. The judicious use of the presentmethodology in the study of breast tumor specimens could givea clear assessment of the cell population dynamics within thetumor and is a unique piece of information that could help predictthe speed of progression for each particular tumor.

If there is one thing that hormone and chemotherapy protocolsfor breast cancer have shown us, it is that in spite of the use ofmultiple drugs, eventually a tumor cell population arises thatescapes the therapy and results in a tumor relapse. The heterogeneity of tumor cell populations could understandably lie atthe basis for this escape. However, heterogeneity of tumor cell

populations per se, as described hitherto, is not enough toexplain their neoplastic evolution, especially since normal breastcell populations are also heterogeneous as shown in this study.As an alternative, the measurement of RPV proposed in thiswork gives an indication of the velocity of change in a heterogeneous cell population. Furthermore, this factor can be quanti-

tated and compared in different normal and tumor cell populations. The RPV could also be valuable in the possible assessmentof genetic predisposition to cancer or in the testing of drugs thatmay reverse the tumor phenotype and slow down metastaticspread.

ACKNOWLEDGMENTS

We wish to thank Johnny Lee for his excellent technical assistance in cell cultureand Marion Douglass for her expert typing.

REFERENCES

1. Breast Cancer Task Force Cell Culture Bank. Current Inventory, Rockville,Md.: EG and G/Mason Research Institute, 1977.

2. Ceriani, R. L, Peterson, J. A., Lee, J. Y., Moneada, R., and Blank, E. W.Characterization of cell surface antigens of normal human mammary epithelialcells with monoclonal antibodies. Somatic Cell Genet., 9. 415-427,1983.

3. Ceriani, R. L., Thompson, K. E., Peterson, J. A., and Abraham, S. Surfacedifferentiation antigens of human mammary epithelial cells carried on the humanmilk fat globule. Proc. Nati. Acad. Sei. U. S. A., 74: 582-586,1977.

4. Gitone, M. A., and Fidler, I. J. Increasing metastatic potential is associatedwith increasing genetic instability of clones isolated from murine neoplasms.Proc. Nati. Acad. Sei. U. S. A., 78: 6949-6952,1981.

5. Curran, R. C., and Gregory, J. Demonstration of immunoglobulin in cryostatand paraffin sections of human tonsil by immunofluorescence and immunoper-oxidase techniques. J. Clin. Pathol. (Lond.), 37: 974-983,1978.

6. Elmore, E., Kakunaga, T., and Barrett, J. C. Comparison of spontaneousmutation rates of normal and chemically transformed human skin fibroblasts.Cancer Res., 43:1650-1655,1983.

7. FÃ¤rber,E. The pathology of experimental liver cell cancer. In: H. M. Cameron,D. A. Linsell, and G. P. Warwick (eds.), Uver Cell Cancer, pp. 243-277.Amsterdam: Elsevier-North Holland Publishing Co., 1976.

8. Farber, E., and Cameron, R. The sequential analysis of cancer development.Adv. Cancer Res., 37:125-226,1980.

9. Fogh, J., and Trempe, G. New human tumor cell lines. In: J. Fogh (ed.), HumanTumor Cells in Vitro, pp. 115-153. New York: Plenum Publishing Corp., 1975.

10. Goldberg, S., and Defendi, V. Increased mutation rates in doubly viral transformed Chinese hamster cells. Somatic Cell Genet., 5: 887-895,1979.

11. Griffin, J. E., Allman, D. R., Durrani, J. L., and Wilson, J. D. Variation in steroid5 (Â»-reducÃaseactivity in cloned human skin fibroblasts. J. Bioi. Chem., 256:3662-3666.1981.

12. Hager, J. C., Russo, J., Ceriani, R. L., Peterson, J. A., Fligiel, S., Jolly, G., andHeppner, G. H. Epithelial characteristics of five subpopulations of a heterogeneous strain BALB/cfC3H mouse mammary tumor. Cancer Res., 41: 1720-

1730,1981.13. Heppner, G. H., Dexter, D. L., De Nuoci, T., Miller, F. R., and Calabresi, P.

Helerogeneily in drug sensitivity among tumor cell subpopulations of a singlemammary tumor. Cancer Res., 38: 3758-3763,1978.

14. Lasfargues, E. Y., and Ozzello, L. Cultivation of human breasl carcinomas. J.Nail. Cancer Insl., 21: 1133-1147,1958.

15. Luria, S. E., and Delbruck, M. Mulalions of bacteria from virus sensilivily lovirus resistance. Genetics, 28: 491-511,1943.

16. Nowell, P. C. The clonal evolution of lumor cell populations. Science (Wash.D.C.), 794:23-28, 1976.

17. Peterson, J. A. Discontinuous variability, in Ihe form of a geomelric progression,of albumin production in hepaloma and hybrid cells. Proc. Nati. Acad. Sci.U. S. A., 77: 2062-2066,1974.

18. Peterson, J. A. Analysis of Ine discontinuous variation in albumin synlhesis alIhe cellular level. Somatic Cell Genel., 5: 641-651,1979.

19. Peterson, J. A. The widespread nature of phenotypic variability in hepalomasand cell lines, in Ihe form of a geomelric series. J. Theor. Biol., 702: 41-53,

1983.20. Peterson, J. A., Bartholomew, J. C., Slampfer, M., and Ceriani, R. L. Analysis

of expression of human mammary epilhelial antigens in normal and malignanlbreasl cells al Ihe single cell level by flow cylofluorimetry. Exp. Cell Biol., 49:1-14,1981.

21. Pitol, H. C., Shires, T. K., Moyer, G., and Garrelt, C. T. Phenotypic variabilityas a manifestation of iranslalional conlrol. In: H. Busch (ed.), Molecular Biologyof Cancer, pp. 523-534. New York: Academic Press, Inc., 1967.

22. Sasaki, M., Peterson, J. A., and Ceriani, R. L. Quantitatton of human mammaryepithelial antigens in cells cultured from normal and cancerous breasf tissues.

SEPTEMBER 1983 4295




In Vitro (Rockville). 17: 150-158, 1981.23. Smith, H. S., Lan, S., Ceriani, R., Hackett, A. J., and Stampfer, M. R. Clonal

proliferation of cultured nonmalignant and malignant human breast epithelia.Cancer Res., 41: 4637-4643, 1981.

24. Soule, H. D., Vazques, J., Long, A., Albert, S., and Brennan, M. A human cellline from a pleural effusion derived from a breast carcinoma. J. Nati. Cancer

Inst., 51: 1409-1413,1973.25. Stampfer, M., Hallowes, R. C., and Hackett, A. J. Growth of normal human

mammary cells in culture. In Vitro (Rockville), 16: 415-425, 1980.26. Young, R. K., Cailleau.R. M., Mackay. B .and Reeves,W. J., Jr. Establishment

of epithelial cell line MDA-MB-157 from metastatic pleural effusion of humanbreast carcinoma. In Vitro (Rockville),9: 239-245, 1974.

_>j

Â¿:

Fig. 1. Colonies of breast epithelialcells stained by immunoperoxidasetechniques for the surface antigen, NPGP.Bright-field microscopy was used in all cases exceptC. A, positive normal breast epithelial cells (4MB);H, negative breast carcinoma cells (MCF-7);C, same as B but phase microscopy; D, mixed colony of SKBR-3 cells; Â£,mixed colony of BT-20 cells, x 580.




1983;43:4291-4296. Cancer Res Jerry A. Peterson, Robert L. Ceriani, Edward W. Blank, et al. Epithelial CellsAntigen Expression in Normal and Cancerous Human Breast Comparison of Rates of Phenotypic Variability in Surface

Updated version

http://cancerres.aacrjournals.org/content/43/9/4291

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/43/9/4291To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



comparison of rates of phenotypic variability in surface antigen … · of appearance of...

Documents