434 GENETICS OF SOMATIC CELLS 



Objections may be raised against the crossing-over hypothesis on the basis that the 

 frequency of Pj versus P 2 variants was highly unequal with some of the tumors studied. 

 This was not true for all tumors. With some, particularly with those originating in the 

 A x A . CA-Fi-hybrid genotype, nearly equal frequencies were obtained in many cases. 

 The asymmetrical tumors may have genetic constitutions that make the expression of one 

 variant less likely than the other, either because of differences in viability, or as a result 

 of the histocompatibility differences in addition to H-2 that separate the strains used, as 

 pointed out previously. Such barriers may be of different strength for the two parental 

 strains, making the expression of one variant type less likely than the other. In fact, a 

 distinct general preference for one of the parental strains characteristic for each F x geno- 

 type that has been studied 7 16 • 724 does appear when surveying a large number of tumors. 



These interpretations have been discussed at some length to illustrate the vicissi- 

 tudes of indirect reasoning and to emphasize even more the necessity of exploring all 

 possibilities to develop methods of genetic transfer between somatic cells. As they stand, 

 the data are best compatible with a genetic change, and, in the opinion of the author, 

 among the various types of genetic changes that may come into consideration, somatic 

 crossing over appears most probable. If this can be confirmed, it would be of consider- 

 able significance, since genetic mapping of somatic cells would be within reach through 

 this process, as pointed out by Pontecorvo, analogous to his impressive studies on 

 Aspergillus. 1011 



Whatever its mechanism, the formation of isoantigenic variants from F 1 tumors 

 has a certain model relationship to some forms of carcinogenesis and tumor progression. 

 It can be viewed as the loss of certain defined isoantigens that normally prevent growth 

 in a given, incompatible, host genotype. Tumor progression and some forms of 

 tumorigenesis (particularly those involving the endocrine system and certain types of 

 chemical carcinogenesis) are characterized by the gradual loss of responsiveness to vari- 

 ous superimposed, growth-controlling mechanisms in the autochthonous host. In a 

 sense, variant formation may be regarded as an experimental model (even though an 

 unnatural, synthetic one) of the deletion theory of carcinogenesis. Bearing the super- 

 ficial nature of this parallel in mind, it is nevertheless interesting to note 722 certain 

 analogies with the rules of tumor progression as formulated by Foulds. 399, 401 Pro- 

 gression occurs independently in different tumors and leads to independent reassortment 

 of different cellular characters — different tumors show an individuality with regard 

 to the occurrence and the frequency of given variants. Different unit characters 

 undergo progression independently of each other — different isoantigenic variants can 

 be obtained from the same tumor. Progression can occur in successive, distinct steps, 

 and so can variant formation. Progression is essentially a one-way process; the per- 

 manence and stability of variant formation is equally irreversible. Whatever its 

 mechanism, variant formation has demonstrated that stable, irreversible, and heritable 

 changes can occur in somatic cells that lead to the loss of cellular components and there- 

 by convey upon the cell a new ability to grow under circumstances in which its progeni- 

 tor would have been checked by a superimposed, systemic mechanism. 



