468 GENETICS OF SOMATIC CELLS 



To study variation, one will have to choose between systems in vivo and in vitro. 

 The approach in vivo will be largely confined to neoplastic cells at present, and the 

 markers available include drug resistance, hormonal dependence, ability to grow in 

 the ascites form, and isoantigens, to mention only a few. The only characters that have 

 a known mechanism of genetic determination and can be localized on the chromo- 

 somal map are the isoantigens of the histocompatibility system, best known in the 

 mouse. This system, notably the histocompatibility-2 system, has also the largest 

 selectivity; and in reconstruction experiments involving artificial mixtures of cells 

 differing at the H-2 locus, it was possible to isolate selectively H-2 compatible cells even 

 if they only represented an extremely small fraction (10 -7 -10 -8 ) of a large H-2- 

 incompatible population. In our experience, systems of resistance to drugs or ascites 

 convertibility have a much lower selectivity (10~ 5 -10 -6 ). Selectivity and known 

 localization on the map gives preference to the isoantigenic markers also as far as the 

 somatic-crossing-over approach is concerned. For genetic-transfer experiments, it is 

 not quite necessary to have a known localization on the map, but a highly selective 

 system appears to be very important even in this case. 



When the approach in vivo is contrasted with the possibilities in vitro, it is obvious 

 that a higher degree of precision and more controlled conditions can be achieved in 

 tissue culture. The isoantigenic systems represent one exception in this respect, 

 however, since one of the main advantages, the high selectivity and specificity of the 

 homograft reaction which serves to concentrate rare variants, has so far not been 

 duplicated in vitro, although it should be possible to do this, at least with types of cells 

 sensitive to cytotoxic, humoral isoantibodies (that is, primarily with lymphatic and 

 bone-marrow cells and their neoplastic derivatives) . As far as other markers are con- 

 cerned, tissue culture will usually provide a more concise method for experimentation, 

 although it must be kept in mind that there is no reason to believe that the population 

 of cells in vitro is representative of the population in vivo; in fact, there is a great deal of 

 evidence to the contrary. Until large-scale, single-cell cloning of animal cells is 

 possible in vitro, with tissues taken directly from the organism in vivo, it appears most 

 conservative to regard in vitro systems merely as models of what might apply to somatic 

 variation in vivo, but without attempting direct extrapolation. 



As to the points raised by Drs. Snell, Lederberg, and Owen, I would hesitate to 

 conclude, at the present stage of our experiments, that we actually deal with somatic 

 crossing over. It is a very probable explanation, but other possibilities must still be 

 kept in mind. It is conceivable, for instance, that D and K could be different parts of 

 the same molecule, in line with Dr. Owen's model, and that D could be a precursor of 

 K. Thus a loss of D would automatically lead to a loss of K also. This argument is 

 weakened somewhat by the fact that there are strains of mice on record, characterized 

 by H-2 combinations which lack D but contain K. Nevertheless, additional markers 

 will be needed if the possibility of somatic crossing over is to be confirmed, and I 

 certainly appreciate the suggestions of Drs. Snell and Lederberg in this regard. 



