Dr. Jane Setlov 
- 2 - 
May 25, 1979 
polyoma DNA Inserts In X or plasmid vectors, and show variously 0, 6, 9% 
tumorigenicity . Because of the small numbers of animals used, exactly re- 
producible results could not be expected here (similar variability is ob- 
served for tumorigenesis by intact polyoma DNA: see Table 5 in Israel et al., 
J. Virol. 2^9:990 (1979)). An accurate, quantitative determination of tunor- 
igenicity is not possible from these data, but it is clear that for both 
monomer- and diner-insert recombinants the tumorigenicity is of the same 
order of magnitude as that of free polyoma DNA. One can conclude that 
insertion into a procaryotic vector does not significantly change the 
tumorigenicity of polyoma viral DNA . Moreover, genetic expression of the 
inserted DNA clearly occurs in animal cells and this indicates the potential 
danger of recombinant molecules containing DNA from any source, not just 
viruses. 
The second and third columns of Table 2 show that cleavage of circular 
polyoma DNA to fora a linear molecule increases its tumorigenicity, and 
the same is true of recombinant molecules, to about the same extent. The 
cleavage methods used produced free linear polyoma DNA in all cases. These 
results again imply that the tumor-inducing activity of recombinant molecules 
is approximately the same as that of free viral DNA, since the increase in 
activity can be attributed in all cases merely to conversion to the linear 
form, rather than to freeing the viral DNA from the DNA of the vector. 
2. The data concerning inoculation of animals with E^. coli containing 
polyoma recombinant DNA (Table 1, p. 8 of ms) show no tumor induction. Two 
points are pertinent: a) The dose of E. coli used was limited by experi- 
mental necessity to contain only 2 x 10^ recombinant molecules, whereas the 
data for naked DNA (Table 10) show that several times this dose was required 
for measurable results under similar conditions, b) The same authors, in 
their previous Science paper (item 1, p. 890, in document 670), pointed out 
a number of possible reasons why E. coll containing infectious recombinant 
DNA was not infectious when administered by injection. The factors they dis- 
cussed are relevant to injection, but not necessarily to other modes of exposure. 
Injection is not a normal route of exposure and it presents a different set of 
challenges, as compared to normal routes. Therefore, experiments performed by 
Inoculation with bacteri a are not a valid test of tne risks entailed in accidental 
exposure of an animal to bacteria containing recombinant DNA . (Injection of 
naked recombinant DNA, on the other hand, i_s a valid test of the final stages 
of access of the recombinant molecule to the animal. The earlier stages must 
then be tested separately, and this remains to be done.) 
If the Committee would like to discuss more extensively the implications 
of the Martin-Roue risk-assessment experiments, I will be happy to participate. 
Yours sincerely, 
Barbara Hatch Rosenberg J 
Associate Member, Sloan-Kettering 
Institute 
Associate Professor, Biochemistry 
Cornell University Medical College 
[ 277 ] 
