MASSACHUSETTS INSTITUTE OF TECHNOLOGY 
77 MASSACHUSETTS AVENUE 
CAMBRIDGE, MASSACHUSETTS 02139 
DEPARTMENT -OF BIOLOGY 
October 16, 1979 
Room 56-721 
PHONE: (617) 253-1000 
Dr. Donald Fredrickson 
Director 
National Institutes of Health 
Bethesda, MD 20014 
Dear Dr. Fredrickson: 
I am writing in response to the proposal before you to reclassify most 
recombinant DNA experiments performed with non-mobilizable plasmids or 
bacteriophage lambda in Eh coli K12 as requiring PI physical containment. 
I strongly support this proposal, partially on the grounds of our now exten- 
sive experience with recombinant DNA plasmids and lambda phages containing 
yeast DNA propagated in _E. coli K12. These experiments have always been 
P2-EK1, so naturally we have used a great variety of EKl-level vectors and 
hosts. Included in our experience is an as yet unpublished risk-assessment 
experiment carried out in collaboration with Hardy Chan, Malcolm Martin, 
and Wally Rowe at N.I.H. in which our bank of yeast hybrids in pMB9/HB101 
(Petes el; al. (1978) Gene 4_:37-49) (an EK1 system) was extensively passaged 
through mice. The universal experience is that any bacterium carrying a hybrid 
plasmid is unstable relative to a bacterium with less of the foreign DNA. 
In the mice mentioned above (as well as in repeated passage in liquid cultures) 
selection appears to occur the result of which is to delete all DNA not directly 
selected for by the presence of tetracycline: i.e., the tet gene carried by 
pMB9 and the plasmid replication origin. All other sequences tend to be lost. 
This comes, of course, as no surprise, since evolution tends in the direction 
of efficiency, not the propagation of sequences useful only to the investigator. 
It is probably worth generalizing further: everyone to my knowledge who 
has extensive dealings with recombinant DNA has found that the biggest practical 
problem with hybrid plasmids and phages is maintenance. Hybrids are almost 
always grown in selective medium; even casual relaxation of selection causes 
rapid loss of the genes of interest. A priori considerations aside, experience 
now clearly provides a significant body of experimental evidence for the pro- 
position that extra sequences are tolerated poorly by Eh coli K12. Whenever 
we grow a cell carrying a recombinant plasmid or phage extensively, even 
being as careful to select as possible, we nevertheless lose all but the 
selected vector portion in time. 
[ 354 ] 
