The analysis of malignancy by cell fusion: I. hybrids between tumour cells and l cell derivatives

We define malignancy as the ability of tumour cells to grow progressively and kill their host. The idea of using cell fusion to facilitate genetic analysis of this phenomenon is an attractive one; but there have so far been few studies of this kind. Barski, Sorieul and Cornefert (Barski, Sorieul & Cornefert, 1961; Barski & Cornefert, 1962) examined hybrid cells that arose spontaneously in mixed cultures of two mouse cell lines, one highly malignant, the other less so. The ability of the hybrid cells to produce tumours resembled that of the more malignant parent. Hybrids between the more malignant of these two parent cells and normal mouse fibroblasts also appeared to be malignant (Scaletta & Ephrussi, 1965); and so did the hybrids that arose spontaneously in mixed cultures of non-malignant mouse cells and mouse cells rendered malignant by infection with polyoma virus (Defendi, Ephrussi, Koprowski & Yoshida, 1967). These latter hybrids continued to produce the T antigen characteristically associated with polyoma virus infection (Defendi et al. 1967; Defendi, Ephrussi & Koprowski, 1964). The conclusion drawn from these studies was that malignancy was a dominant character in somatic cell hybrids and that therefore the lesion resulting in malignant behaviour was unlikely to involve a loss of genetic information. Silagi (1967) studied the behzviour of hybrids between cells of a malignant melanoma and A 9 cells (a mouse fibroblast line lacking the enzyme inosinic acid pyrophosphorylase). The in vivo behaviour of these hybrids was less clear cut. Some clones produced tumours in a majority of the animals injected; others produced few tumours or none at all. In all of these studies relatively large inocula of cells were used and little attention appears to have been paid to the possibility that the tumours resulting from the injection of the hybrid cells might have arisen from variants in the hybrid cell population that did not reflect the overall level of malignancy of the population as a whole. Since injection of cells into the animal selects for malignancy, the conclusion that malignancy is a dominant characteristic in hybrid cells must remain precarious unless one can exclude the possibility that the tumours produced result from the selective outgrowth of a minor subpopulation of the hybrid cells injected. This hazard applies, of course, to any cell population injected into the animal; but it is especially serious in the case of somatic cell hybrids, because these hybrids are known to exhibit progressive loss of chromosomes and other forms of chromosomal instability (for review, see Harris, 1970). These cells therefore present an unusually high degree of genetic variation on which selection could operate. Barski & Cornefert (1962) and Ruddle, Chen, Shows & Silagi (1970) have examined the chromosomal constitution of some cell populations explanted from tumours produced by the injection of hybrid cells; but, with the possible exception of a single tumour examined by Silagi (1967), no information appears to have been published about the chromosomal constitution of the tumours themselves. It is therefore difficult to assess the extent to which previous studies indicating a dominance of malignancy in somatic cell hybrids might have been complicated by the loss of chromosomes or by selection of atypical variants in vivo. The introduction of the Sendai virus cell fusion technique (Okada, 1958; Harris & Watkins, 1965) has made it possible to hybridize virtually any mammalian cells and has thus greatly increased the range and power of this form of analysis. It therefore seemed worthwhile, in view of the great biological and clinical importance of malignancy, to re-investigate these questions in a more systematic way. The present series of papers deal with the growth in vivo of hybrids produced by fusing a range of highly malignant mouse cells with cells of established lines of low tumorigenicity and with diploid cells isolated directly from the animal. Preliminary accounts of some of this work have appeared previously (Harris et al. 1969; Harris & Klein, 1969); and a detailed analysis of the immunological characteristics of the hybrid cells and the tumours arising from them has been published elsewhere (Klein, Gars & Harris, 1970).

Cells of the following ascites tumours were used: Ehrlich, SEWA, MSWBS, YAC and YACIR. The Ehrlich tumour was originally derived at about the turn of the century from a mammary carcinoma in a mouse of unknown genetic constitution. This tumour has been passaged continuously in mice of different genotypes and appears to have evolved mechanisms that greatly reduce the expression of its histocompatibility antigens (Hauschka & Amos, 1957). It grows in any strain of mouse and produces lethal neoplasms when injected subcutaneously as well as intraperitoneally: a few cells injected into the peritoneal cavity will kill most mice within 3 weeks; even one cell is lethal in about 15 % of recipient animals (Hauschka, 1953). The SEWA tumour is an ascites sarcoma originally induced in 1960 by the injection of polyoma virus subcutaneously into a newborn A.SW mouse (Sjögren, Hellstrom & Klein, 1961). This tumour carries the H-2^a histocompatibility antigen complex and its growth is limited to mice of the A. SW strain. It also carries the polyoma-specific transplantation antigen. Although originally an osteogenic sarcoma, the tumour now shows no osteogenic differentiation. An inoculum of 1000 cells produces neoplasms in 100% of genetically compatible unirradiated mice, and an inoculum of 100 cells produces neoplasms in a 100% of mice given 400 rd (4 J kg^{− 1}) of total body X-irradiation (Sjöigren, 1964). The SEWA tumour was converted to the ascites form in 1968 (Nordenskjöld, 1968), and, in this form, an inoculum of 10⁶ cells kills most animals in 3-4 weeks. Like SEWA, the MSWBS tumour is an ascites sarcoma. It also carries the H-2^a histocompatibility antigen complex and is specific for the A.SW strain of mice. It was originally derived from a tumour induced by the injection of methylcholanthrene into the thigh muscle of a male A. SW x AF₁ hybrid mouse (Klein & Klein, 1958). The variant used in the present experiments arose from cells passaged in the parental A.SW strain. This variant has lost the H-2^a antigen complex derived from the A strain parent, but has retained the H-2^a complex of the A.SW strain parent. Growth of the tumour is restricted to A.SW mice. An inoculum of as few as 50 cells produces neoplasms in 100% of unirradiated genetically compatible mice. The neoplasms are lethal within 14-20 days. The YAC and YACIR tumours are 2 different ascitic sublines of a lymphoma induced in a mouse of the A/Sn strain by Moloney virus (Klein, Klein & Haughton, 1966). The YACIR subline was derived from YAC by serial passage in strain A hosts immunized against the surface antigen associated with Moloney virus (Fenyö, Klein, Klein & Swiech, 1968). YACIR shows a 10-fold reduction in the concentration of the Moloney surface antigen, compared with YAC, and is resistant to the cytotoxic action of anti-Moloney antisera. With inocula of about 10 cells, both YAC and YACIR sublines produce neoplasms in 100% of strain A mice, but the cells are rejected in other strains of mice. With inocula of 10 cells, the neoplasms produced are usually fatal within 10–14 days.

The established cell lines used (A9, B82 and A9RI) were all derivatives of the L cell (Earle, 1943). The A9 cell was obtained by progressive selection for resistance to 8-azaguanine (Little-field, 1964a) and lacks the enzyme inosinic acid pyrophosphorylase, a defect that prevents its growth when de novo synthesis of purines is blocked by aminopterin (Littlefield, 1964b). The B82 cell was obtained by selection for resistance to bromodeoxyuridine and lacks the enzyme thymidine kinase (Littlefield, 1966). The A9RI cell line was derived from a single colony of cells that arose in bulk cultures of A9 cells exposed to HAT medium (see below). This medium prevents the growth of A9 cells, which lack inosinic acid pyrophosphorylase, but not of cells that possess this enzyme. This single colony of cells, which survived the treatment of the A9 cultures with HAT medium, proved on further subculture to be refractory to the inhibitory effects of this medium and to contain a high level of inosinic acid pyrophosphorylase. The inosinic acid pyrophosphorylase activity of the L cell is approximately 100 units per unit protein, that of the A9 cell is barely detectable and that of the A9 RI cell is approximately 60 units per unit protein. Chromosomal analysis of the A9 RI cell showed a karyotype essentially similar to that of the A9 cell, but the modal number of chromosomes was slightly lower than that of the wild type A9 cultures. The chromosomal constitution of the various parental cell lines is given in Table 1. The A9, B82 and A9RI cells all bore the H-2^k histocompatibility antigen complex. This is consistent with their derivation from L cells which were originally obtained from C₃H mice carrying the H-2^k histocompatibility complex.

The cells were grown as monolayers in plastic flasks (Falcon Plastics, Los Angeles, California, U.S.A.) or glass bottles in Dulbecco’s medium (Vogt & Dulbecco, 1963) or HAT medium. HAT medium was originally devised by Szybalska & Szybalski (1962) to select against cells lacking inosinic acid pyrophosphorylase. It is composed essentially of Eagle’s minimum essential medium (Eagle et al. 1956) with added aminopterin, hypoxanthine, thy midine and glycine at concentrations of 4 × 10^{− 7} M, 1 × 10^{− 1} M, 1.6 × 10^{− 6} M and 3 × 10^{− 6} M respectively. All media contained foetal calf serum at a concentration of 10 % and the following antibiotics: penicillin 200 μg, streptomycin 200 μg kanamycin 30 μg neomycin 30 μg and mycostattin 40 μg/ml.

The cells were fused together by means of inactivated Sendai virus as described by Harris & Watkins (1965). The Ehrlich and SEWA cells fused readily with the L cell derivatives, but the MSWBS cells fused rather less well; the YAC and YACIR lymphomas fused poorly. Adequate numbers of heterokaryons could, however, be formed from the more recalcitrant cell types by increasing the concentration of Sendai virus and the relative input of the less fusable cell types (Harris, Watkins, Ford & Schoefl, 1966).

Ehrlich, SEWA, YAC and YACIR cells adhere poorly to glass or plastic and may be eliminated simply by frequent changes of medium. MSWBS cells do adhere, but they are easily detached and they grow rather poorly on the plastic surface, at least initially. Selection against the tumour cells thus presented little difficulty. A9 and B82 cells do not grow in HAT medium, so that when these cells are fused with the tumour cells and the virus-treated cell population is grown in HAT medium, only hybrid cells, in which the tumour cells complement the enzymic defects of the A9 or B82 cells, survive.

Selection of hybrids between the Ehrlich cells and the A9RI cells presented a more difficult problem, since growth of the A9RI cells, which have a high level of inosinic acid pyrophosphorylase, is not inhibited by HAT medium. In this case, differences between the adhesive properties of the A9RI cells and the hybrids between these cells and the Ehrlich cells were exploited to select for the hybrids. Whereas the Ehrlich cells did not adhere to the plastic or glass surfaces of the culture vessels, the Ehrlich/A9RI hybrids did adhere to these surfaces, but much less firmly than the A9RI cells themselves. The hybrid cells were thus more easily detached from the floor of the flasks than the A9RI cells and they frequently came away spontaneously at mitosis. The cultures were therefore passaged by shaking the contents vigorously and transferring the resulting cell suspension to fresh flasks. This procedure led to progressive enrichment of the cultures with hybrid cells and, after several passages, to the production of pure cultures of hybrid cells.

Metaphase spreads of ascites tumours and of cells growing in vitro were made by the air drying technique of Rothfels & Siminovitch (1958). The cells were exposed to colcemid (Ciba Pharmaceutical Products Inc.) at a concentration of 0.1 μg/ml for 1–2 h, treated with 0.9% sodium citrate solution for about 10 min and then fixed in 1:3 acetic acid-methanol. The fixed cells were spread on wet slides, air dried and stained with buffered Giemsa solution. Chromosome preparations from solid tumours were made in the following way. The mice bearing the tumours were given a dose of colcemid equivalent to 8–10 μg/g of body weight and then killed 6–8 h later. Small pieces of the tumour were passed through a fine steel gauze mesh in 0.9 % sodium citrate solution or in tissue culture medium diluted 1:5 with water. This procedure took 30-40 min. A fine suspension of cells was separated from the tissue mince by slow centrifugation and this cell suspension was then fixed and stained in the same way as the cells of the ascites tumours.

Except where one of the parents was the Ehrlich cell, all hybrids were assayed in syngeneic F₁ mice: the SEWA/A9 and MSWBS/A9 hybrids in A. SW × C3H, the YAC/A9 and YACIR/A9 hybrids in A× C3H. The tumour cells grow as well in the F₁ hybrid test animals as in the parental strains. The original genotype of the mouse in which the Ehrlich tumour arose is not known, so that assays of the hybrids between the Ehrlich cells and the L cell derivatives were carried out in C3H mice, from which the L cell was derived. The Ehrlich tumour grows perfectly well in C3H mice, as in all other strains. In general, the histocompatibility antigen complexes of both the parent cells were expressed in the hybrids, although in hybrids in which one of the parents was an Ehrlich cell, the histocompatibility antigens of the other parent were expressed poorly (Klein et al. 1970). The possibility remained, however, that minor degrees of histoincompatibility might have been generated between the long established cell lines and the animals from which they were initially derived; and cell fusion might itself have given rise to new, undetected, antigenic combinations, even though direct tests on the hybrids showed the presence of both the expected parental antigen complexes. Assays for tumorigenicity were therefore carried out on newborn mice some of which were given 4 J kg^{− 1} of whole body X-irradiation. Since newborn mice given this dose of radiation commonly support the growth of tumours bearing foreign H-z antigens, it seemed unlikely that this assay would be complicated to any important degree by histoincompatibility between the hybrid cells and the test animals. In the event, this expectation was fulfilled. As will be seen later, when tests in syngeneic irradiated newborn mice revealed differences in rumorigenicity among the different clones of the one hybrid cell type, these differences were also apparent when the cells were inoculated into allogeneic irradiated newborn mice. In the tests the cells were injected subcutaneously in inocula of between 4 × 10⁴ and 3.3 × 10⁶ cells. Small cohorts of animals were injected from time to time, as cells and appropriate F₁ hosts became available. The inoculated animals were examined for development of tumours at weekly intervals. No animal was scored as negative until a period of 3 months had elapsed, and many groups of animals were observed for 5-6 months. In selected groups, growth of the tumours was recorded by regular caliper measurements: 3 diameters were measured and the geometric mean calculated.

Although some of the tumours produced in test animals by the injection of the hybrid cells could not be successfully passaged, others proved to be transplantable. These were carried in unirradiated adult mice syngeneic with the animal in which the tumour originally arose. Solid tumours were passaged by the subcutaneous injection of 0-2 ml of a crude cell suspension. Tumours converted to the ascites form were passaged by intraperitoneal injection of 02 ml of undiluted ascitic fluid. Occasional ascites tumours arose directly from the injection of the hybrid cells intraperitoneally into unirradiated adult mice.

This was carried out by P. R. Cook using the method described by Harris & Cook (1969).

It will be seen from Table 2 that A9 cells and B82 cells have a very limited ability to grow progressively in vivo in irradiated newborn C3H mice. Even with very large inocula these cells rarely give rise to progressive tumours. The A9RI cell produces tumours more readily than the A9 cell; but it still has a very low level of tumorigenicity in comparison with the various tumours used in the present study.

Clones of cells derived from these 2 parents showed great variation in cell shape, clonal morphology and growth rate. Tight epithelial clones, clones composed of long spindle-shaped cells growing in parallel, highly dispersed clones of irregular fibroblastic cells and a whole range of intermediate categories were observed. The in vivo tests were carried out on the most vigorous hybrids, which soon overgrew the bulk cultures. These hybrids grew rapidly in vitro with a generation time of between 10 and 12 h. The hybrid nature of the cells was confirmed by karyotypic analysis (Table 3). Both the Ehrlich cell and the A9 cell have specific chromosomal markers (Harris et al. 1965; Engel, McGee & Harris, 1969). The markers of both parent cells were identified in the hybrids. The hybrid cells initially contained a modal number of 128 chromosomes of which approximately 24 were bi-armed. This is very close to what one would expect from the fusion of one modal Ehrlich cell and one modal A9 cell. The cells showed little change in chromosome number during the first 2 months in culture, but, on more prolonged cultivation, chromosomes were progressively eliminated. As shown in Table 4, the Ehrlich/A9 hybrids had a very low level of tumorigenicity, comparable to that seen in the A9 cell. With inocula of up to 3·5 × 10⁸ cells, the cumulative take incidence was only about 10%, even in irradiated newborn animals. Because these hybrids were selected in medium containing high concentrations of aminopterin, thymidine and hypoxanthine, their ability to produce tumours was also tested after the cells had been adapted to growth in medium not containing these additives. The take incidence was not increased. It is clear that the highly malignant character of the Ehrlich cell was not transmitted to the hybrid cell.

Although the tumorigenicity of these hybrids was very low, occasional solid and ascites tumours did develop after variable latent periods of 3-12 weeks. A few of these tumours proved to be transplantable. When, however, the chromosomes of these tumours were examined, the cells in the tumours were found to have a much lower modal chromosome number than that of the hybrid cells originally injected into the animal. Whereas the initial hybrid cell population had, at the time of injection, a modal chromosome number of about 128, the tumours produced had modal chromosome numbers in the eighties (Table 5). That the cells of which the tumours were composed were derived from hybrid cells was, however, clear from the presence in them of both Ehrlich and A9 chromosomal markers. The tumours were thus produced, not by the continued progressive growth in vivo of the hybrid cell population as a whole, but by selective overgrowth of variants which had lost chromosomes. This form of selection in the progressive growth of tumours is, of course, well known (for review see Berger, 1969). The results thus appeared to indicate that Ehrlich/A9 hybrids containing the complete chromosomal complements of both parents showed little or no capacity for progressive growth in vivo, for even when the injection of such hybrids did produce occasional tumours, these were not composed of cells with unreduced chromosome complements.

These results raise several obvious questions. (1) Is the suppression of malignancy by cell fusion limited to the Ehrlich ascites tumour? (2) Is the effect peculiar to the A9 cell? (3) Does the metabolic defect in the A9 cell (inosinic acid pyrophosphorylase deficiency) play an important role in the suppression of malignancy? (4) Can the suppressive effect be exercised by normal diploid cells? (5) Is it, in general, the case that when a malignant cell and a non-malignant cell are fused together, the hybrids with unreduced chromosome complements are not malignant? (6) Is a loss of chromosomes essential for the production of malignant variants from a non-malignant hybrid cell population ? (7) If loss of chromosomes is essential, is it necessary to eliminate certain specific chromosomes, or is it enough simply to achieve some overall reduction in chromosome number? In the experiments now to be described attempts were made to investigate each of the above questions.

The hybrid nature of all these cell lines was confirmed by karyotypic analysis and by the presence on the surface of the cells of both the parental sets of H-2 antigens (Harris et al. 1969; Klein et al. 1970; unpublished observations by Eva-Maria Fenyo & Gertrud Grundner). The chromosome constitutions of the hybrid lines are shown in Table 3. The modal chromosome number of the MSWBS/A9 hybrid was initially almost exactly the sum of the modal chromosome numbers of the 2 parent cells. The modes of the SEWA/A9, YAC/A9 and YACIR/A9 hybrids were a little lower than those to be expected from the fusion of 2 modal parent cells. Marker chromosomes characteristic of the A9 cell were identified in all hybrids. SEWA and MSWBS markers were also identified in the SEWA/A9 and the MSWBS/A9 hybrids; but for the YAC/A9 and YACIR/A9 hybrids, karyological identification rested on the presence of the A9 markers, the total number of chromosomes and the proportion of bi-armed chromosomes, since the YAC and YACIR tumours lack specific chromosomal markers. The identification of these hybrids was, however, confirmed by the detection on the surface of the cells of the H-2 antigen complexes of both parents (observations by Eva-Maria Fenyo & Gertrud Grundner). The chromosomal constitutions of all these hybrids were initially very stable in vitro. There was little change in the modal chromosome number over the first weeks of cultivation, but, after several months, progressive loss of chromosomes was observed (Table 3). The stability of the karyo-type in these hybrids in vitro resembled that previously described for a wide range of intraspecific hybrid cells (for review see Harris, 1970.)

The growth of these hybrids in vivo is shown in Table 4. It will be seen that in all cases the malignancy of the tumour cell was profoundly modified by fusion with the A9 cell. All these hybrids had a very low level of malignancy compared with that of the tumour cell parent. The take incidences of the SEWA/A9 and MSWBS/A9 hybrids were slightly higher than that of the Ehrlich/A9 hybrids; but the YAC/A9 and YACIR/A9 hybrids produced very few tumours. It is clear that the A9 cell has the ability to suppress the malignancy not only of the Ehrlich ascites cell, but also a wide range of other tumour cells.

The chromosomal constitutions of some of the tumours produced by the SEWA/A9 and MSWBS/A9 hybrids and of all the tumours produced by the YAC/A9 and YACIR/A9 hybrids were examined. In all cases the cells in the tumours showed a marked reduction in modal chromosome number relative to that of the hybrid cell population injected. As in the case of the Ehrlich/A9 hybrids, the tumours arose, not by progressive growth of the hybrid cell population as a whole, but by selective overgrowth of cells that had lost chromosomes.

The question arises whether the malignancy of these tumour cells might not be suppressed simply by continued cultivation in vitro. One subline of the MSWBS tumour and 2 sublines of the SEWA tumour were therefore adapted to growth in vitro and cultivated continuously for several months under the same conditions as the hybrid cells. The ability of these tumour cells cultivated in vitro to grow progressively in vivo was then tested. The results of the tests are shown in Table 6. Prolonged cultivation in vitro has clearly not suppressed the malignancy of these tumour lines.

In order to test whether the ability to suppress malignancy was limited to the A9 cell, hybrids were made between the Ehrlich cell and the B82 cell, an L cell derivative lacking thymidine kinase. These hybrids initially had a slightly lower modal chromosome number than the Ehrlich/A9, hybrids (Table 3). The take incidence for the Ehrlich/B 82 hybrids is shown in Table 4: it is little different from that of the Ehrlich/A9 hybrids. The B82 cells are thus no less effective than the A9 cells in suppressing the malignancy of the Ehrlich tumour cells.

Both the A9 and the B82 cells are, however, cells with severe metabolic defects. Although the Ehrlich cells have a very high level of both inosinic acid pyrophosphorylase and thymidine kinase, it is possible that other metabolic derangements associated with inosinic acid pyrophosphorylase deficiency (Felix & DeMars, 1969), or thymidine kinase deficiency, might limit the growth of the hybrid cells in vivo. The Ehrlich cell was therefore fused with the A9RI cell, apparently a revertant of the A9 cell in which inosinic acid pyrophosphorylase activity is restored. (The inosinic acid pyrophosphorylase level in the A9 RI cell is about 60 units per unit protein, compared with about 100 units for the L cell.) The take incidence for the Ehrlich/A9 RI hybrids (Table 4) was indeed somewhat higher than that for the Ehrlich/A9 hybrids, but the level of tumorigenicity was still very low compared with that of the Ehrlich tumour cells. It was therefore clear that the ability of the A9 cell to suppress malignancy was not simply a consequence of its inosinic acid pyrophosphorylase deficiency, although it was possible, since the take incidence of the Ehrlich/A9RI hybrid was higher than that of the Ehrlich/A9 hybrid, that the inosinic acid pyrophosphorylase deficiency of the A9 cell might have been a contributory factor. It is, however, very unlikely that the growth of the hybrid cells in vivo is limited by low inosinic acid pyrophosphorylase levels. The take incidence of the Ehrlich/A9 RI hybrids, despite the contribution from the Ehrlich cell, was actually lower than that of the A9 RI cells themselves (Tables 2, 4); and analysis of some ascitic variants of transplantable tumours arising from Ehrlich/A9 hybrids showed no correlation between inosinic acid pyrophosphorylase content and growth rate.

The most striking feature of the tumours produced by the hybrids tested in the present investigation was the gross reduction in the modal chromosome number of the tumour cells compared with that of the hybrid cells injected. This indicated that in these hybrid cell populations only cells that had lost chromosomes were capable of progressive growth in vivo, or, at least, that the growth of such cells was very much more rapid in vivo than that of the unreduced hybrids. If the generation of malignant variants from an essentially non-malignant hybrid cell population required simply an overall reduction in chromosome number and not the loss of certain specific chromosomes, one might expect that the tumorigenicity of the hybrid cell population would increase if the cells were grown in vitro long enough to permit substantial chromosome loss to occur. The Ehrlich/A9, hybrids were therefore grown in vitro for a period of about 18 months during which the modal chromosome number dropped from about 128 to less than 80 (Table 3), the latter figure being comparable to the modes found in the various tumours derived from Ehrlich/A9, hybrids (Table 5). However, as shown in Table 4, the take incidence for these hybrids with grossly reduced chromosome numbers was not higher than that obtained before substantial loss of chromosomes had occurred. This finding indicates that the generation of malignant variants in the nonmalignant hybrid cell population requires the loss of specific chromosomes and not simply an overall reduction in chromosome number. It appears that injection of the cells into the animal selects for just those hybrids that have lost the specific chromosomes responsible for suppression of the malignancy; but growth in vitro does not select for these hybrids, so that their incidence in the population is not greatly increased. Analysis of chromosome losses in intraspecific mouse cell hybrids cannot provide conclusive evidence for this interpretation, but the information that can be obtained is entirely consistent with it. After 18 months’ cultivation in vitro, when the modal chromosome number of the Ehrlich/A9, hybrid cells had fallen from 128 to less than 80, about two-thirds of the bi-armed chromosomes derived from the A9 cell were retained; but in the tumours derived from these hybrids a comparable reduction in chromosome number was often associated with preferential elimination of the bi-armed A9 chromosomes, and in some tumours all but one or two of these chromosomes were eventually eliminated (Table 7). This does not, of course, mean that the chromosomes responsible for the suppression of malignancy were necessarily the bi-armed A9 chromosomes. Indeed, the studies of Ruddle et al. (1970) on tumours derived from hybrids between malignant melanoma cells and A9 cells suggest that this is not so. These tumours appeared to be composed of cells in which the bi-armed A9 chromosomes were retained, although the observations were subject to some uncertainty because the chromosomal analyses were not done on the tumours themselves, but on populations of cells explanted from the tumours and grown in vitro before analysis. It is, however, clear that the chromosomes eliminated from the cells that grow progressively in vivo are not the same as those lost on continued cultivation of the cells in vitro.

The present findings show that the ability of a wide range of different malignant cells to grow progressively in vivo can be suppressed when these cells are fused with certain derivatives of the L cell line. The highly malignant character of the tumour cells is suppressed whether or not the L cell derivatives have metabolic defects that facilitate selection of the hybrid cells. So long as these hybrid cells retain the complete chromosome complements of the two parent cells, their ability to grow progressively in vivo is very limited, for tumours composed of such unreduced hybrids have not been found. However, when they lose certain specific, but as yet unidentified, chromosomes, the hybrid cells may regain the ability to grow progressively in vivo and give rise to a tumour. These results indicate that the L cell derivatives contribute something to the hybrid that suppresses the malignancy of the tumour cell, and this contribution is lost when certain specific chromosomes are eliminated. We have now to see whether this suppressive effect can also be produced by normal diploid cells and, if so, whether in this case also, reversion to malignancy is associated with loss of chromosomes.