Proc. Nat. Acad. Sci. USA Vol. 71, No. 5, pp. 2062-2066, May 1974 Discontinuous Variability, in the Form of a Geometric Progression, of Albumin Production in Hepatoma and Hybrid Cells (reiterated genes/control of gene expression and differentiation) JERRY A. PETERSON* Centre de Gbnbtique Molkculaire, C.N.R.S., 91, Gif-sur-Yvette, France Communicated by Daniel Mazia, February 7, 1974 ABSTRACT A clonal rat hepatoma cell line (Fu5) activation of a gene(s) that was repressed during differentia- produces rat serum albumin at a constant rate over at least tion. These results have been confirmed in other interspecific 3 months of continuous cultivation. Ten hybrid cell clones derived from the fusion of Fu5 cells and mouse fibroblasts, hybrid cells (6, 7). The reappearance in hybrids of enzymes as well as 14 hepatoma subclones of Fu5 cells, all produce that were lost during cultivation in vitro of one of the parental albumin but at different rates, ranging from about 0.09 to cell lines has also been observed (8, 9). 36.7 Mg/mg of protein per 72 hr. Despite this variability in- The present paper deals with the other example of pheno- albumin production, the distribution of clones is not in and hybrid cells, namely a random but discontinuous, with both hepatoma and typic variability hepatoma hybrid clones clustering around discrete values that can be variation in constitutive level of albumin production. fitted to the geometric progression: a, a(V/2)1, a(V22)2..... MATERIALS AND METHODS a(V/2)". The values of the majority of clones fall into alternate members of this geometric progression, which Cells were cultured in a modified Ham's F12 medium supple- differ by a factor of 2. Hepatoma subclones with indis- mented with 5% fetal-calf serum, as described (5). tinguishable karyotypes differ in level of albumin pro- The origins of cell lines Fu5 and 3T3 and the isolation of the duction by as much as 4-fold. In contrast to hepatoma used in these studies have been described (10). All clones, albumin production in hybrid clones decreases hybrids with increasing cell generations. A survey of 28 enzymes of other hepatoma clones are subclones of Fu5. different hepatomas reveals a large variability in enzyme To assay for the production of rat serum albumin (RSA), 2 levels which, for most enzymes, can be arranged into X 106 hepatoma, 1.5 X 106 3T3, or 1 X 106 hybrid cells were classes that form a geometric progression. The apparent seeded in 250-ml Falcon plastic tissue-culture bottles con- widespread nature of this discontinuous phenoLypic variability suggests that it may reflect a basic mechanism taining 15 ml of medium. For each tell type, the number of of control of gene expression in animal cells. cells seeded was such that they just approached confluency after 72 hr of growth. After an incubation period of 72 hr, the Phenotypic changes occur in animal cells in culture at a much medium was removed. The cells were detached from the bottle higher frequency than would be expected as a result of muta- with trypsin, collected, and counted, and an aliquot was tion, i.e., a change in nucleotide sequence of the structural washed once with phosphate-buffered saline (11) and then gene. The rate of mutation in bacteria is about 10-9-10- per frozen for protein assay at a later date by the method of cell per generation, while studies with mouse myeloma cells Lowry (12). reveal a higher rate of phenotypic change (10-3 per cell Rat albumin was assayed by microcomplement (C') fixa- per generation) (1). Studies of the effect of ploidy on rate of tion (13) with a total reaction volume of 7 ml. All reac- "mutation" in hamster (2) and frog (3) cells are also incom- tants were diluted in the standard C' fixation buffer (10 mM patible with changes in nucleotide sequences being involved. Tris-HCi, pH 7.4; 0.15 M NaCl; 0.15 mM CaCl2; 0.5 mM Phenotypic variability is also observed in neoplasia. Not MgSO4; and 0.1% gelatin w/v). Rabbit antiserum against only do different tumors arising from the same tissue differ in rat albumin was prepared from rabbits immunized with puri- their degree of malignancy, but they also have different mor- fied rat albumin as described (5). One antiserum from a single phological, biochemical, and growth properties (4). Since bleeding that had good complement fixation activity was used many phenotypic changes observed in both tumors and cell throughout. Maximum C' fixation was obtained with 0.085 lines are apparently not genetic, they may reflect a perturba- ,ug of purified rat albumin per reaction tube (Fig. 1) and this tion of normal cellular mechanisms controlling cell differentia- was used as a standard for estimating albumin content in tion, for differentiation is generally presumed to involve selec- growth media. This purified rat albumin is shown to be identi- tive gene expression rather than irreversible changes in genetic cal to that in freshly prepared rat serum since the two C' fixa- information. Thus, the study of phenotypic variability could tion curves superimpose completely (14). The rat serum dilu- help to elucidate the control of differentiation and neoplasia. tion that gives maximum C' fixation is 1: 400,000, equivalent My studies on the production of serum albumin in rat to 0.085 Asg of albumin. Samples of medium to be assayed were hepatoma cells and rat hepatoma-mouse fibroblast hybrids centrifuged at 500 X g for 5 min, heated at 600 for 20 min, reveal two novel examples of phenotypic variability. One is dialyzed against two changes of 100 times the sample volume the induction of mouse albumin production in rat hepatoma- of C' fixation buffer without gelatin, centrifuged at 27,000 X mouse fibroblast hybrids (5) and the other is the subject of g for 30 min, and then frozen until assayed. Dialysis of the this paper. Since the mouse fibroblasts did not produce mouse medium samples allowed addition of up to 4 ml of sample to albumin, its production by the hybrids must result from the the reaction mixture rather than 1 ml as in the original technique, thus increasing the sensitivity of the assay 4-fold. *Present address: Laboratory of Radiobiology, University of Control experiments with purified rat albumin diluted in California, San Francisco, Calif. 94143. dialyzed growth medium gave a peak of C' fixation at the Abbreviations: C', complement; RSA, rat-serum albumin. same albumin concentration as when it was diluted in C' fixa- 2062 Downloaded by guest on September 27, 2021 Proc. Nat. Acad. Sci. USA 71 (1974) Discontinuous Variability of Albumin Production 2063 TABLE 1. Rat albumin production in __. hepatoma-fibroblast hybrids 80 No. of 68c Hybrid Aug/mg of protein gener- Class clone per 72 hrt P* ations -'40 3F12 2.95 <0.01 25 20: 8 3F11 2.90 (3.04,2.76) <0.01 33 4 a.01 0.05 0.1 0.5 110 2.9 0.08t [2.5] pg RSA/be 6 3F13 1.33 <0.02 27 Ic. d. 3F14 1.28 (1.17,1.4) <0.02 26 1.3 A 0.07$ [1.25] 3F15 0.85 (0.74,0.94) n.s. 26 4 3F16 0.66 (0.64,0.68) <0.05 26 '40 3F10 0.78 <0.05 52 4 0.76 i 0.05$ [0.625] 20 20/ 2 3Fla 0.37 (0.38,0.37) [0.31] <0.05 42 ~dj 0.01 0.05 0.1 0.5 1.0 0.0 0.05 0.1 0.5 1.0 0 3FCa§ -0.12 85 pg RSA/fbe pg PSA/tt 3F75m§ -0.09 FIG. 1. C' fixation curves obtained with antibody against rat albumin, diluted 1:1800; and rat albumin (RSA) (0); (b) * Probability that the rate of production is the same as the medium from hepatoma (A), hybrid (0), or fibroblast (V) cells; average rate of the closest adjacent class. (d) rat albumin diluted in growth medium (X); or growth t Values in brackets are of a theoretical geometric progression medium alone (E). (a and c): standard curves for the two inde- (see Results). dependent assays shown in (b) and (d), respectively. t Mean of class i SEM. § Amount of albumin produced is at lower limit of assay. highest producing clones is more than 20-fold; however, the n.s. = not significant. hybrids are not distributed randomly but cluster around discrete values, which can be fitted to a geometric progression tion buffer. Standard deviation for repeated assays of a single where successive group values differ by a factor of 2 (Table 1). medium collection and for assays of media collected on dif- Since there is some loss of chromosomes in all of the hybrid ferent days from a given clone were similar, between 10 and clones (7-16%) (10), it is possible that the respective levels 16% when expressed per mg of cell protein. Samples frozen simply represent the number of chromosomes carrying the for as long as 6 months showed no significant change in con- albumin gene present in the particular hybrid. This is an un- tent of rat albumin detectable by C' fixation. likely interpretation since it would not explain the hybrid classes forming a geometric progression. Moreover, there is no RESULTS correlation between the gross karyotype (10) of the hybrids Ten hybrid cell clones derived from the fusion of rat hepatoma and the level of albumin they produce. cells (clone Fu5) with 3T3 mouse fibroblasts secrete rat-serum The observation that the hybrids that have gone through albumin (RSA) into their growth medium.