Suppression of Malignancy in Human Cancer Cells: Issues and ChallengesAuthor(s): Albert B. SabinSource: Proceedings of the National Academy of Sciences of the United States of America,Vol. 78, No. 11, [Part 2: Biological Sciences] (Nov., 1981), pp. 7129-7133Published by: National Academy of SciencesStable URL: http://www.jstor.org/stable/11423 .

Accessed: 01/05/2014 22:35

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact support@jstor.org.

.

National Academy of Sciences is collaborating with JSTOR to digitize, preserve and extend access toProceedings of the National Academy of Sciences of the United States of America.

http://www.jstor.org

This content downloaded from 62.122.73.193 on Thu, 1 May 2014 22:35:57 PMAll use subject to JSTOR Terms and Conditions

Proc. Natl. Acad. Sci. USA Vol. 78, No. 11, pp. 7129-7133, November 1981 Medical Sciences

Suppression of malignancy in hum Issues and challenges ALBERT B. SABIN

Medical University of South Carolina, Charleston, South Carolina 29425

Contributed by Albert B. Sabin, July 20, 1981

ABSTRACT Analysis of the many, sometimes seemingly con- tradictory, reports on the partial suppression of malignancy in highly unstable rodent intraspecies and rodent-human hybrid cells emphasizes the limitations of this approach to the analysis of the basic nature of malignancy, especially in naturally occurring human cancers. During the past 5 years, Stanbridge and then Klinger reported complete suppression, not elimination, of malig- nancy [defined as capacity to produce progressively growing tu- mors in athymic (nude) mice] in stable hybrids of different human cancer cells with normal human fibroblasts or with differentiating epithelial keratinocytes and, importantly, also in stable hybrids of two parental cancers of different somatic cell origin. The nontu- morigenic human hybrid cells are not rejected by some nonthymic immune mechanism of nude mice and survive in vascularized foci; the initial multiplication of these cells is stopped by some unknown proliferation controlling substance(s) to which their malignant par- ent(s) do not respond. The heritable properties of infinite multi- plication in vitro, loss of contact inhibition, etc. remained in the nontumorigenic hybrids but, remarkably, the in vitro production of a human choriogonadotropin by HeLa cells was suppressed along with tumorigenicity and reappeared in the tumorigenic re- vertants. If it is-assumed that human cancers.of different somatic cell origin are caused by a loss of different specific regulatory genes, as the most recent data reviewed here suggest, the chal- lenge is to determine in molecular terms what those missing genes are, how they function, and whether it may be possible to restore to the cancer cells what they have lost.

The complete suppression of malignancy in stable hybrids of human cancer cells with normal human fibroblasts first reported by Stanbridge (1) and confirmed and extended by Klinger et al. (2, 3) has opened new pathways for research on the nature and perhaps also specific therapy of human cancers. The purposes of this communication are (i) to review previous work on the analysis of malignancy by cell fusion and to evaluate recent data on stable complete suppression of malignancy in human cancer cells by hybridization with normal human cells or with human cancer cells of different somatic cell origin, (ii) to call attention to the possibly important information that could emerge from a resolution of seemingly contradictory results reported by dif- ferent investigators, and (iii) to indicate some of the many chal- lenges that intraspecies human cell hybrids present for joint research among experimental oncologists, cytogeneticists, and molecular biologists to elucidate the nature of the malignant process in different human cancers by direct studies on natu- rally occurring human cancer cells rather than, as is so often done now, on experimental cancers that may bear little or no relationship to the great variety of human cancers.

Because transformation in vitro is not synonymous with tu-. morigenicity in vivo (4-6), in the following discussion I use "malignancy" to refer only to the property of progressive tu- morigenicity in vivo. Thus, "malignancy" relates to the capacity of a cell for continued multiplication in vivo when no more of

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

712'

an cancer cells:

it is needed and in parts of the body where it does not belong. One of the important issues in the genesis of human cancers is whether this attribute is the result of.acquisition of totally new genetic information (as in experimental cancers produced by viruses or by large doses of chemical carcinogens or radiation that result in extensive DNA damage followed by faulty repair) or of a mutational malfunction or loss of preexisting specific reg- ulatory suppressor genes (in the one cell from which the tumor arises) which prevent normal cells from behaving postnatally as they did during early embryonic development. Unstable suppression of malignancy in rodent intraspecies and rodent-human interspecies hybrids The elaborate studies on malignancy of hybrid cells of varying parentage that followed the new techniques of in vitro cell fu- sion developed by Harris and Watkins (7) were lucidly sum- marized in several publications (8-10). The initial report of the collaborative studies by Harris and coworkers (11) provided the first evidence for what they called "suppression of malignancy by cell fusion." However, with some exceptions, what has been called "suppression of malignancy" in intraspecies or interspe- cies rodent hybrids has not been a complete absence of tumor- igenicity of the cultured hybrid cells but rather a graded char- acter that was measured by the number of cells required to produce progressively growing tumors in irradiated immuno- logically permissive histocompatible mice-i.e., much larger numbers of the hybrid cells than of the highly malignant parent cells had to be injected in order to obtain progressively growing tumors in 50-100%' of the mice. Similar findings were also re- ported shortly thereafter by Ephrussi et al. (12).

The need for much larger numbers of initially cultured hybrid cells to produce tumors that consisted of hybrid cells was in- terpreted as indicating that malignancy was completely sup- pressed in the hybrid cells with a full or nearly full complement of chromosomes from both parental cells and that the tumors resulted from the in vivo selection of the hybrid cells that had lost certain chromosomes and were just as malignant as the highly malignant parent (8, 11). Because there was great vari- ation in the total number of chromosomes that were found in the hybrid tumor cells, it was postulated that reversion to ma- lignancy depended not on the loss of a special number of chro- mosomes but rather on the loss of certain specific chromosomes which carried the hypothetical normal suppressor genes (13). Based on this interpretation of the observed gradation of ma- lignancy in various hybrid cell populations, Harris (8) proposed the following postulate: "If the malignant phenotype were due to some form of genetic loss, or the synthesis of some nonfunc- tional gene product, then one would expect that this defect would be made good by a normal cell in which the relevant gene or genes are unimpaired. Complementation between the ma- lignant and the normal cell would then be expected to restore the normal phenotype. Moreover, if more than one type of ge- netic loss, or impaired gene function, could determine a ma-

Abbreviations: ahCG, a human choriogonadotropin; NK, natural killer cells; SV40, simian virus 40.

This content downloaded from 62.122.73.193 on Thu, 1 May 2014 22:35:57 PMAll use subject to JSTOR Terms and Conditions

7130 Medical Sciences: Sabin

lignant phenotype it should be possible in some cases to show complementation between different kinds of malignant cell: hybrid cells of nonmalignant phenotype might then be produced from two malignant cells"[italics mine].

It is noteworthy that the quantitative malignancy suppressive effects referred to above were obtained with naturally occurring mouse tumors as well as with mouse tumors experimentally produced by viruses or chemicals (8-11). These results with mouse tumors of different origin and the absence of tumori- genicity, in nude mice, of human cells experimentally trans- formed by the Epstein-Barr virus and simian virus 40 (SV40) (4) led Klein (6) to suggest that, whereas certain viruses and chemical carcinogens can transform cells to become "immortal" and to exhibit other abnormal properties in vitro, they are not per se responsible for tumorigenicity in vivo, which appears to depend on additional cellular changes that result in "resistance to negative feedback regulation of the host."

The reports of Croce, Koprowski, and Aden (14, 15) that in- corporation of the SV40 genome can by itself confer tumori- genicity because malignancy was not suppressed in hybrids of SV40-transformed human cells with normal mouse cells, which retain a full complement of mouse chromosomes and only a few human chromosomes or even a single human chromosome, are not in accord with the above postulate, even if one assumes that normal mouse cells transformed by SV40 would regularly pro- duce progressively growing tumors in nude mice. Using irra- diated, newborn histocompatible mice instead of nude mice in their tests for tumorigenicity, Gee and Harris (5) found that not only were SV40-transformed mouse fibroblasts usually not tu- morigenic but also such transformed mouse cells were able to suppress tumorigenicity completely in hybrids with a highly malignant mouse melanoma; moreover, they also demonstrated that in tumorigenic SV40-transformed mouse cells it was pos- sible to suppress this tumorigenicity by hybridization with nor- mal mouse fibroblasts. Gee and Harris (5) were inclined to at- tribute the discrepant data of Croce, Koprowski, and Aden (14, 15) to their use of 107-108 cells in their tests for tumorigenicity because the small number of tumors resulting from these large numbers of hybrid cells could have been due to an in vivo se- lection of rare malignant variants. This may or may not be the whole explanation, and the assumption that naturally occurring cancers and experimental cancers artificially produced by vi- ruses or chemicals have the same ultimate defects responsible for malignancy needs to be resolved by further work. For ex- ample, the DNA of chemically transformed malignant mouse and rat cells has recently been reported to transform normal mouse cells to malignancy (16), in the presence of normal "sup- pressor genes" unless they were destroyed by the manipulation required to get the DNA into the cells-a possibility that can and should be tested.

The capacity of cultured mouse embryo fibroblasts (at 2-4 x 106 cells) to suppress malignancy completely in some clones of hybrids with a highly malignant mouse melanoma was dem- onstrated in well-controlled studies by Jonasson et al. (17). Ac- cordingly, it is especially meaningful (and potentially highly in- formative) that Halaban et al. (18) found no suppression of malignancy (at least at 1 X 106 cells per mouse, the only dosage tested) in clones of another malignant mouse melanoma with normal mouse melanocytes that appeared spontaneously in tu- mors produced in newborn mice or experimentally after in vitro fusion with skin cells of i-day-old mice. In one such hybrid clone the full complement of parental chromosomes was present in the cultured cells before they were inoculated into mice and in the tumors these cells produced. It is noteworthy that the nor- mal, nonmalignant parents of these hybrids were differentiating melanocytes and not fibroblasts, which raises the important question whether normal differentiating cells may have no func-

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

tioning malignancy suppressor genes at the time they are undergoing differentiation.

More difficult to interpret, however, is the apparent absence of suppression in spontaneous in vivo hybrids with a full com- plement of parental chromosomes that appeared in mouse fi- brosarcoma tumors in 3-month-old syngeneic mice (19).

In 1978, Koprowski et al. (20) reported that a hybrid of a tu- morigenic human melanoma with nontumorigenic mouse fi- broblasts was tumorigenic in nude mice even when only human chromosomes 14, 17, and 21 remained. One year later, Carney et al. (21) reported that 14 independent clones of hybrids of tu- morigenic human lung cancer cells and nontumorigenic mouse embryofibroblasts were completely nontumorigenic in 77 nude mice (each inoculated with 4-12 X 106 hybrid cells). The pres- ence of many human chromosomes in these hybrids was estab- lished by isoenzyme markers. Is it possible that the human chro- mosomes carrying the derepressed genes for the human lung cancer were lost in all the clones tested by Carney et al. (21) and that the few human chromosomes that remained in the melanoma hybrid cells tested by Koprowski et al. (20) retained the derepressed genes responsible for the malignancy of the melanoma, and that normal mouse fibroblasts cannot suppress human malignancies because "malignancy suppressor genes" are species-specific? Or is it possible that the normal mouse fibroblasts used by Carney et al. (21) contained "malignancy suppressor genes" while those of Koprowski et al. (20) did not?

The same issue arises in hybrids between malignant mouse tumors and normal human fibroblasts. In 1977, Jonasson and Harris (22) reported that hybrids of a highly malignant mouse melanoma and their line of normal human fibroblasts were non- tumorigenic in nude mice even in a clone in which only one human chromosome (an X chromosome) was left. Two years later, Kucherlapati and Shin (23) reported the opposite for hy- brids of the same mouse melanoma and also of another mouse cancer with two different lines of normal human fibroblasts (but they did not test the lines used by Jonasson and Harris). They found that 19 independently arising clones were uniformly tu- morigenic in nude mice (although more hybrid cells than ma- lignant parent cells were required to produce tumors), despite the demonstration by cytological and isoenzyme analysis that at least 14 of the 23 human chromosomes were retained both in the original hybrids and in the progressively growing tumors in nude mice. Here again, one faces the issue of possible species specificity of the suppressor gene(s) or that the remaining nor- mal human chromosomes did not carry the required comple- ment of suppressor genes.

The extensive and detailed chromosomal and tumorigenicity studies on many cloned hybrids derived from "spontaneous" malignant mouse or Chinese hamster tumors and different nor- mal human fibroblasts by Klinger and his associates (2, 24, 25) indicated that "no human chromosome in single copy appears to be able to greatly affect tumorigenicity" but "many hybrids, particularly those with many human chromosomes, require higher cell inocula or longer periods of time to form tumors than do the heteroploid parental lines." Moreover, their data also showed that certain combinations of chromosomes from the normal human cells were never found in tumors produced by the hybrid cells, and one or more combinations were always present in the suppressed hybrids. The most effective sup- pressor combinations appeared to be the normal human chro- mosomes 9 and 17,'11 and 13, and 11 and 17. These observations do not account for the unusual finding by Jonasson and Harris (22), but they led Klinger to the tentative conclusion "that at least two mutations are required for a cell to undergo malignant transformation" (25) and that "genes on at least two chromo- somes appear to be required for suppression" (24).

Quite different results were reported by Bloch-Shtacher and

This content downloaded from 62.122.73.193 on Thu, 1 May 2014 22:35:57 PMAll use subject to JSTOR Terms and Conditions

Medical Sciences: Sabin

Sachs (26) from work with malignant SV40-transformed or methylcholanthrene-transformed Chinese hamster cells; they concluded: "The results indicate that chromosome 3 carries gene(s) that control malignancy in Chinese hamster cells in cell lines transformed by a viral or a chemical carcinogen and that malignancy was induced in both cell types by an increase of these genes." Is it possible that the mechanism of malignancy is different in cancers experimentally produced by viruses or chemicals than in naturally occurring cancers, despite the con- clusions to the contrary (6) based on the early experiments of Harris, Klein, and their associates (9, 1I)? It is also noteworthy that Howell and Sager (27) found evidence for a role of cyto- plasmic genes in the partial suppression of malignancy in a spon- taneous Chinese hamster tumor but not in an experimentally produced SV40 mouse tumor. However, working with a spon- taneous mouse mammary carcinoma, Ziegler (28) also found no evidence for any role of cytoplasmic transfer of suppression of malignancy and concluded: "The expression of tumorigenicity in cybrids is consistent with the idea that the controlling mech- anisms for malignant potential reside in the nucleus of somatic cell hybrids." These seeming differences in malignancy suppression factors in different species also emphasize the im- portance of analyzing the nature of malignancy in different hu- man cancers by direct studies on human cancer cells.

Unique properties of human carcinoma-human normal fibroblast hybrids with stable, complete suppression of malignancy From heteroploid HeLa cells (50-64 chromosomes; modal, 61) which produced progressively growing tumors in 100% of nude mice inoculated with 105 cells, Stanbridge et al. (1, 29-31) cre- ated hybrids with senescent or actively proliferating human diploid fibroblasts (45-47 chromosomes; modal, 46) which pro- duced no tumors when 5 x 107 cells were inoculated in each of many immunosuppressed or nude mice. When cultured in vitrofor 4-6 weeks after fusion, the hybrids initially lost a small number of chromosomes (from an expected 95-111 to 86-104) but failed to produce tumors in immunosuppressed or nude mice even after inoculation of 107 cells per mouse subcutane- ously, intraperitoneally, intramuscularly, intrathoracically, or intracerebrally (5 x 105 cells) (1, 29).' Thus, even individual cells that had only 86-90 chromosomes were not selected for tu- morigenicity in vivo. During subsequent cultivation over ape- riod of 2 years, there was no further loss of chromosomes, and the hybrid cell lines remained nontumorigenic. However, the nontumorigenic hybrids, which morphologically were inter- mediate between the epithelial malignant cells and the fibro- blastic normal parents, retained the following in vitro properties of transformation exhibited by the malignant but not the normal parent cells: (i) capacity to multiply indefinitely (i.e., "immor- tality"); (ii) absence of contact inhibition; (iii) anchorage-inde- pendent growth (colonies in soft agar); and (iv) lectin agglutin- ation, decreased requirement for serum factors, alterations in expressions of various cell surface components, and production of alkaline phosphatase. However, the capacity of the HeLa cells to produce a human choriogonadotropin (ahCG) in vitro was lost by the nontumorigenic hybrids by (E. J. Stanbridge, personal communication, work with H. Sussman).

After many months of continuous cultivation, rare morpho- logically altered subpopulations were isolated from three HeLa-fibroblast hybrid lines, and these formed progressively growing tumors in nude mice inoculated with 5 x 10 cells (31, 32). These tumorigenic revertants, which had lost <5% of the total.biparental chromosome complement (32), were morpho- logically like the HeLa cells and had regained the capacity to produce ahCG in vitro (E. J. Stanbridge, personal communi- cation). This linkage between in vitro production of ahCG and

Proc. Natl. Acad. Sci. USA 78.(1981) 7131

tumorigenicity is of special interest because either the same genes are responsible for both or different genes close together on the same chromosome(s) are absent from or inactive in the tumorigenic cell and can be suppressed by active genes in nor- mal cells. Chromosomal analysis of the tumorigenic revertant cells showed "that the loss of one copy* of chromosomes 11 and 14, respectively, is correlated with a high degree of statistical significance, to the expression of tumorigenicity" (32). This finding is of special interest because of the previous demon- stration by Klinger et al. (2, 24, 25) of the suppressive role of normal human chromosome 11 (although only in combination with chromosome 13 or 17) in interspecies hybrids with mouse and hamster malignant tumors.

All the phenomena.reported by Stanbridge et al. were fully confirmed and extended by Klinger et al. (2, 3) who did not test for ahCG production. They used a clonal derivative from the same line of HeLa cells used by Stanbridge but different normal human cells (male fetal fibroblasts) which had nine distinctive chromosomes that could be identified in the hybrids. All their hybrids remained completely nontumorigenic after subcloning and prolonged culture. They showed that nontumorigenic hy- brids were also obtained even when HeLa cells derived from a nude mouse tumor were used for hybridization, thus elimi- nating the possibility of hybrids being formed preferentially from a subpopulation of nontumorigenic cells. By recovering rare tumorigenic variants from cloned populations of nontu- morigenic hybrids, they also showed that tumorigenicity in the hybrids was only suppressed and not eliminated by the hybrid- ization procedures. These revertant cells had lost only a small number of chromosomes, primarily those of the distinguishable normal parent chromosomes,t which led Klinger (3) to con- clude: "only specific chromosomes can carry such [suppressive] information, since other isolates with equally low chromosome numbers were not tumorigenic."

Validity of assumption that tumorigenicity in nude mice is an indicator of malignancy instead of immunologic rejection by nonthymic mechanisms Because nude mice have been shown to have high levels of nat- ural cell-mediated cytotoxic or natural killer (NK) activity which has been reported to inhibit proliferation of some mouse lym- phoid tumors (33, 34), and because the suitability of nude mice as indicators of malignancy has been recently questioned (35-37), Stanbridge and Ceredig (38) carried out an' extensive study on their nontumorigenic hybrids and the revertant tu- morigenic segregants derived from them, which showed the following.

(i) During the first 3 days after injection, there was no his- tologic difference between the nontumorigenic and tumori- genic cell foci; within 24 hr, both showed extensive central ne- crosis surrounded by a rim of intact cells with many mitotic figures.

(ii) On day 4, mitotic activity ceased in the nontumorigenic cells which assumed a fibroblastoid morphology and did not multiply further; the focus of remaining cells was vascularized, there was no inflammatory response or other cellular infiltra- tion, and the cells were alive as was evident by their multipli, cation in vitro.

(iii) Mitotic activity remained very active in the tumorigenic cells, and progressively growing tumors developed.

(iv) Nontumorigenic hybrid cells produced no tumors in neo- natal or heavily irradiated nude mice which lack NK cells and other immune mechanisms.

* The three remaining copies may have been from the malignant parent in which the suppressor genes had been lost or nonfunctional.

t Data.as to which normal chromosomes were lacking in the revertants are not yet available.

This content downloaded from 62.122.73.193 on Thu, 1 May 2014 22:35:57 PMAll use subject to JSTOR Terms and Conditions

7132 Medical Sciences: Sabin

(v) Nontumorigenic hybrids and the tumorigenic segregants were equally susceptible to the cytotoxic action of NK cells of nude mice in vitro, but when NK cell activity was induced in nude mice by C. parvum vaccine, the tumorigenic segregants still formed tumors.

(vi) Tumor-derived hybrid cells were also sensitive to NK cell toxicity in vitro, indicating that the tumors are not made up of cells that are selected for resistance to NK cell toxicity.

(vii) Injection of a mixture consisting of 90% nontumorigenic hybrid cells and 10% tumorigenic cells did not prevent tumor formation, indicating that the nontumorigenic hybrid cells did not induce a nonthymic immune response to the tumorigenic segregants derived from them.

These experiments indicate that these human cancer cells that are tumorigenic in nude mice have lost the something that permits a response to the proliferation inhibitory substance(s) produced by nude mice in vivo which stopped the proliferation of their nontumorigenic hybrid clonal relatives. The nontu- morigenic hybrid cells have regained the genes responsible for this feedback response property of normal cells, despite the fact that they still had the capacity for infinite multiplication in vi- tro-i. e., immortality. Whether or not the in vitro transformed properties of immortality, absence of contact inhibition, etc. that have been found by Stanbridge et al. (1, 29) and Klinger (3) to be present in all the nontumorigenic hybrid cells represent necessary preliminary stages ("precancerous") in the progres- sion to malignancy, it would appear that the mechanism of the production of these heritable properties is different from that involved in the ultimate production of malignancy which can be completely suppressed by hybridization with normal cells while the transformation properties detected in vitro except for the production of ahCG are not suppressed. Can malignancy of other human cancers also be suppressed by hybridization with normal human fibroblasts? According to Stanbridge (personal communication), hybrids of human lung carcinoma (A549) and of endometrial carcinoma cells, having modal chromosome numbers of 63 and 51, re- spectively, with normal human fibroblasts have also yielded nontumorigenic hybirds. However, special problems were en- countered in hybridization tests with the human sarcoma HT1080 which has cells with pseudodiploid and tetraploid num- bers of chromosomes. Stanbridge found that, when pseudodi- ploid clones of HT1080 were used for fusion with normal fibro- blasts, the hybrids usually did not proliferate sufficiently in vitro to provide enough cells for tumorigenicity tests. However, one such hybrid did proliferate enough and its tumorigenicity was suppressed. Hybrids with tetraploid clones of HT1080 hu- man sarcoma proliferated adequately in vitro but tumorigenic- ity was not suppressed. Croce et al. (39), also working with HT1080 human sarcoma and different lines of human diploid fibroblasts, succeeded in obtaining proliferating hybrids with pseudodiploid sarcoma cells; these hybrid cells, which appar- ently contained the full complement of chromosomes from the sarcoma and normal cells, were uniformly tumorigenic when 5 X 106 or 107 cells were injected in nude mice, although they also counted nonprogressive nodules >5 mm in diameter with- out further growth during 2 or more months as tumors. They reported that "karyologic analysis of the tumorigenic hybrid cells recovered from the tumors indicated that these cells were either near tetraploid or near hexaploid human hybrid cells." This is a problem that obviously needs to be resolved, especially the failure of normal fibroblasts to achieve suppression of the tetraploid HT1080 human sarcoma.

Complementation in hybrids derived from human cancers of different somatic cell origin Harris (8) postulated that, if hybrids of cells from two different malignant tumors could be shown to be nontumorigenic, it

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

would indicate that "more than one kind of genetic loss, or im- paired gene function, could determine a malignant phenotype" and that the unaffected genes of one complemented the lost or impaired genes of the other. Such complementation has now been definitely demonstrated in the stable hybrids of human tumors of different somatic cell origins by B. E. Weissman and E. J. Stanbridge (personal communication). They found that there was no evidence of suppression in carcinoma-carcinoma and carcinoma-B-cell lymphoma hybrids, but there was evi- dence of suppression in stable carcinoma-sarcoma and carci- noma-melanoma hybrids; however, most of the latter hybrid clones showed a chromosomal instability not previously en- countered in human tumor-normal human fibroblast hybrids. The human carcinomas in the above tests included two different lines of cervical carcinoma, lung carcinoma, metastatic adeno- carcinoma, and teratocarcinoma. The B-cell lymphomas were one Epstein-Barr virus-carrying Burkitt lymphoma and another B-cell lymphoma. The fibrosarcoma (HTD114) was a pseudo- diploid clone of the HT1080 sarcoma discussed above, and it is noteworthy that tumorigenicity was completely suppressed in the two clones of the lung carcinoma-sarcoma hybrids.

Tumorigenicity of malignant SV40-transformed human fibroblast-normal human fibroblast hybrids It is now well established that the presence of SV40 genome in transformed human (4) or mouse (5) cells is not by itself suffi- cient to make the cells tumorigenic. Suppression of malignancy in a tumorigenic SV40-transformed mouse fibroblast line by hybridization with normal mouse fibroblasts has been reported (5). One group of investigators (14, 15) has continued to report that a human SV40-transformed fibroblast line of low tumori- genicity in nude mice (10s cells producing only occasional, non- progressive tumors) not only is not suppressed by hybridization with normal mouse cells but, in at least one instance, even re- sulted in hybrids of greater tumorigenicity-i. e., 107 cells pro- duced large progressively growing tumors in 100% of nude mice (15). The question, already discussed earlier, is whether or not there may be a difference in the mechanisms of malignancy in human cells, in which viruses produced transformation prior to the ultimate change(s) leading to experimental tumorigenicity, and in naturally occurring human cancers. B. E. Weissman and E. J. Stanbridge (ref. 40; personal communication) hybridized an SV40-transformed tumorigenic human fibroblast line (106 cells produced progressively growing tumors in 10 of 10 nude mice) with normal human fibroblasts, and "a full spectrum, ranging from complete suppression (one hybrid with a modal number of 90 chromosomes) to full expression of tumorigenicity (one hybrid with a modal number of 101 chromosomes) includ- ing hybrid clones (with modal chromosome numbers of 98-105) that were less tumorigenic than the parental SV40-transformed cell line (modal number of chromosomes of 67)." These results with SV40-tumorigenic fibroblast-normal fibroblast hybrids are different from the suppression oftumorigenicity regularly found in numerous clones of the naturally occurring tumorigenic ep- ithelial (HeLa) cell-normal human fibroblast hybrids, but in view of the tumorigenicity encountered in the naturally occur- ring fibroblastic HT1080 sarcoma-normal human fibroblast hy- brids, I conclude only that more work is needed.

Behavior of human undifferentiated carcinoma (HeLa)-normal differentiating human keratinocyte hybrids Primary cultures of normal epithelial epidermal keratinocytes can differentiate in vitro and, when 106-107 cells are inoculated subcutaneously in nude mice, they rapidly form differentiated cystic structures that do not progress to detectable nodules (41). Peehl and Stanbridge (41) produced many hybrids of such pri- mary epithelial cells with HeLa cells and found, first of all, that the capacity of the HeLa cells to form progressively growing

This content downloaded from 62.122.73.193 on Thu, 1 May 2014 22:35:57 PMAll use subject to JSTOR Terms and Conditions

Medical Sciences: Sabin

tumors in nude mice was completely suppressed in the hybrids. However,initially or after many subsequent population dou- blings, some of the hybrid clones produced nonprogressive nod- ules only a few millimeters in diameter at some of the subcu- taneous sites in which 106-107 hybrid cells were inoculated. Histologically, these nodules had the appearance of differen- tiated squamous cell carcinomas with few or no dividing cells. According to Peehl and Stanbridge "the key question is whether these nodules represent bonafide tumors or whether they are the result of a growth regulatory signal mediated by the host which switches the dividing HeLa-keratinocyte hybrids into a differentiated, non-dividing state." In any case, these nonma- lignant hybrids provide still another opportunity for analysis of malignancy in human cancer cells.

Some challenges The work already done during the past 5 years by both Stan- bridge and Klinger and their associates on the complete suppression of malignancy in stable hybrids of different human cancers with normal cells and also with cancers of different so- matic cell origin have provided strong evidence for the concept that malignancy in human cancer cells is the result of a loss or malfunction of different specific regulatory genes that are pres- ent in normal somatic cells. The use of such stable, nontumori- genic hybrids provides hitherto unexplored possibilities for re- search directly on human cancers. Here are only a few of the many challenges that come to mind.

1. Use available technology (42, 43) to determine whether certain metaphase chromosomes from normal fibroblasts can be incorporated into HeLa cells and which ones or how many are needed to suppress their malignancy. Changes in morphology and failure to produce ahCG in vitro could serve as screens.

2. Use the same technique to determine whether the trans- fer of certain chromosomes from HeLa cells to normal fibro- blasts and nontumorigenic hybrids will induce malignancy in one or the other or both, or not at all.

3. Determine whether transfer of malignancy to normal hu- man fibroblasts can be achieved with sonicated DNA from HeLa cells as it was done with DNA from methylcholanthrene-in- duced mouse cancer cells, spontaneous mouse, and human car- cinoma to normal mouse fibroblasts (16, 44).

4. Determine whether sonicated DNA from normal human fibroblasts can suppress malignancy of HeLa cells. If so, it would be possible to use restriction enzymes to determine which piece(s) carry the postulated suppressor genes.

5. How do suppressor genes in normal chromosomes that are present (but not as alleles) in the nuclei of nontumorigenic hy- brids suppress the malignancy expressed by the genes in the derepressed chromosomes from the malignant parent? If sup- pressor RNA or protein can be identified, explore possibilities of preparing it in large amounts and of getting it into cancer cells during the early stages of progressive growth in nude mice.

6. Test many more varieties of human cancer for their ca- pacity to be suppressed by hybridization with normal human cells, and determine the role of naturally integrated viral ge- nome in some ofthem-for example, can the malignancy in cells from primary hepatocellular carcinoma with integrated hepa- titis B virus genome be suppressed as well as that in cells with- out this genome, assuming that suppression can be achieved at all with these cancer cells?

7. In view of the remarkable parallelism between suppression of malignancy in vivo and suppression of production of ahCG in vitro in hybrids of HeLa cells and normal human cells, de- termine whether ahCG may be the substance that is respon- sible for the failure of the malignant HeLa cells to respond to the in vivo proliferation-inhibiting factor(s) by testing the effect of anti-ahCG agents on the progressive multiplication of HeLa cells in nude mice.

Proc. Natl. Acad. Sci. USA 78 (1981) 7133

1. Stanbridge, E. J. (1976) Nature (London) 260, 17-20. 2. Klinger, H. P., Baim, A. S., Eun, C. K., Shows, T. B. & Ruddle,

F. H. (1978) Cytogenet. Cell Genet. 22, 245-249. 3. Klinger, H. P. (1980) Cytogenet. Cell Genet. 27, 254-266. 4. Stiles, G. D., Desmond, W., Jr., Sato, G. & Saier, M. H., Jr.

(1975) Proc. Natl. Acad. Sci. USA 72, 4971-4975. 5. Gee, C. J. & Harris, H. (1979)J. Cell Sci. 36, 223-240. 6. Klein, G. (1979) Proc. Natl. Acad. Sci. USA 76, 2442-2446. 7. Harris, H. & Watkins, J. F. (1965) Nature (London) 205,

640-646. 8. Harris, H. (1971) Proc. R. Soc. London. Ser. B 179, 1-20. 9. Miller, O. J. (1974) in Chromosomes and Cancer, ed. German,

J. (Wiley, New York), pp. 521-563. 10. Ozer, H. L. & Jha, K. N. (1977) Adv. Cancer Res. 25, 53-92. 11. Harris, H., Miller, O. J., Klein, G., Worst, P. & Tachibana, T.

(1969) Nature (London) 223, 363-368. 12. Ephrussi, B., Davidson, R. L. & Weiss, M. C. (1969) Nature

(London) 224, 1314-1315. 13. Klein, G., Bregula, U., Wiener, F. & Harris, H. (1971)J. Cell

Sci. 8, 659-672. 14. Croce, C. M., Aden, D. & Koprowski, H. (1975) Proc. Natl. Acad.

Sci. USA 72, 1397-1400. 15. Koprowski, H. & Croce, C. M. (1977) Proc. Natl. Acad. Sci. USA

74, 1142-1146. 16. Shih, C., Shilo, B. Z., Goldfarb, M. P., Dannenberg, A. & Wein-

berg, R. A. (1979) Proc. Natl. Acad. Sci. USA 76, 5714-5718. 17. Jonasson, J., Povey, S. & Harris, H. (1977) J. Cell Sci. 24,

217-254. 18. Halaban, R., Nordlund, J., Francke, U., Moelmann, G. & Ei-

senstadt, J. M. (1980) Somat. Cell Genet. 6, 29-44. 19. Aviles, D., Jami, J., Rousset, J.-P. & Ritz, E. (1977)J. Natl. Can-

cer Inst. 58, 1391-1397. 20. Koprowski, H., Steplewski, Z., Herlyn, D. & Herlyn, M. (1978)

Proc. Natl. Acad. Sci. USA 75, 3405-3409. 21. Carney, D. N., Edgell, C. J., Gazdar, A. F. & Minna, J. D.

(1979) J. Natl. Cancer Inst. 62, 411-415. 22. Jonasson, J. & Harris, H. (1977)J. Cell Sci. 24, 255-263. 23. Kucherlapati, R. & Shin, S. (1979) Cell 16, 639-648. 24. Klinger, H. P. & Shows, T. B. (1979) Cytogenet. Cell Genet. 25,

172-173. 25. Klinger, H. P. (1981) in Chromosomes Today, eds. Ford, C. E.

Bennett, M. D., Babrow, M. & Hewitt, G. (Allen & Unwin, Lon- don), Vol. 7, pp. 220-224.

26. Bloch-Shtacher, N. & Sachs, L. (1977) J. Cell. Physiol. 93, 205-212.

27. Howell, A. N. & Sager, R. (1978) Proc. Natl. Acad. Sci. USA 75, 2358-2362.

28. Ziegler, M. L. (1978) Somat. Cell Genet. 4, 477-489. 29. Stanbridge, E. J. & Wilkinson, J. (1978) Proc. Natl. Acad. Sci.

USA 75, 1466-1469. 30. Der, C. J. & Stanbridge, E. J. (1978) Cell 15, 1241-1251. 31. Der, C. J. & Stanbridge, E. J. (1980) Int.J. Cancer 26, 451-459. 32. Stanbridge, E. J., Flandermeyer, R., Daniels, D. & Nelson-

Rees, W. (1981) Somat. Cell Genet., in press. 33. Kiessling, R., Petranyi, G., Klein, G. & Wigzell, H. (1976) Int.

J. Cancer 17, 275-281. 34. Warner, N. L., Woodruff, M. F. A. & Burton, P. C. (1977) Int.

J. Cancer 20, 146-155. 35. Kahn, P. & Shin, S. (1979) J. Cell Biol. 82, 1-16. 36. Reid, L. M., Jones, C. L. & Holland, J. (1979)J. Gen. Virol. 42,

609-614. 37. Minato, N., Bloom, B. R., Jones, C., Holland, J. & Reid, L. M.

(1979)J. Exp. Med. 149, 1117-1133. 38. Stanbridge, E. J. & Ceredig, R. L. (1981) Cancer Res. 41,

573-580. 39. Croce, C. M., Barrick, J., Linnenbach, A. & Kporowski, H.

(1979)J. Cell. Physiol. 99, 279-285. 40. Stanbridge, E. J., Der, C. J., Doersen, C-J., Nishimi, R. Y.,

Peehl, D. M., Weissman, B. E. & Wilkinson, J. E. (1981) Sci- ence, in press.

41. Peehl, D. M. & Stanbridge, E. J. (1981) Int. J. Cancer 27, 625-632.

42. Cassingena, R., Suarez, H. G., Lavialle, C., Persuy, M. A. & Ermonval, M. (1978) Gene 4, 337-349.

43. Persuy, M. A., Ermonval, M., Suarez, H. G., Daya-Grosjean, L., Nardeux, P., Giovanella, B. C., Kay, A. C. & Cassingena, R. (1980) Exp. Cell Biol. 48, 51-65.

44. Shih, C., Padhy, L. C., Murray, M. & Weinberg, R. A. (1981) Nature (London) 290, 261-264.

This content downloaded from 62.122.73.193 on Thu, 1 May 2014 22:35:57 PMAll use subject to JSTOR Terms and Conditions