55106 
Federal Register / Vol. 47, No. 235 / Tuesday, December 7, 1902 / Notices 
to one hundred thousand-fold less than 
the parent plasmids from which they 
were derived.) 
Risk assessment studies were carried 
out in volunteers to quantitate the 
mobilizability of pUK325 in comparison 
with p|HK5. Fifty billion E. coli I IS cells 
containing p|I3K5. which encodes 
tetracycline resistance, were fed to 15 
volunteers who were also given 
tetracycline. All 15 individuals became 
heavily colonized by E. coli t IS. and 
excreted IIS along with indigenous 
normal coliforms. Within 24 hours, 9 of 
the 15 individuals had evidence of p|nK5 
in their normal flora (transconjugants) 
indicating that transfer of the plasmid 
had occurred. None of 12 volunteers 
who ingested E. coliHS carrying poorly- 
mobilizable plasmid pI3R325 had 
transconjugants, despite colonization by 
1 IS in the presence of normal flora and 
despite ingestion of tetracycline. 
E. coli I IS containing both pBR325 and 
conjugative plasmid F-amp was fed to 
two groups of volunteers. One group 
took tetracycline daily, the other did not. 
Of 18 subjects who were colonized with 
HS carrying both pDR325 and F-amp and 
who took tetracycline, normal flora 
transconjugants carrying pBR325 were 
found in 13 individuals. In contrast, none 
of the eight volunteers who were 
colonized with HS carrying both pBR325 
and F-amp and who did not take 
tetracycline, showed transconjugants. 
These studies demonstrate that the 
poorly mobilizable plasmid pBR325 
could be mobilized in vivo only under 
extremely contrived conditions, i.e., the 
plasmid resided in a colonizing strain of 
E. coli (1 IS) along with a conjugative 
plasmid, 5X10 10 /?. coli HS were 
ingested with buffer, and volunteers 
received tetracycline daily. Under the 
NIH Guidelines for Research Involving 
Recombinant DNA Molecules, pBR325 
would reside in a weakened strain of E. 
coli (K— 12) without a co-residing 
conjugative plasmid, and laboratory 
workers on antibiotictherapy would be 
evaluated by the institution to determine 
whether they should be engaged in 
recombinant DNA research. In addition, 
the chance for an accidental ingestion of 
a large inoculum (5X10 10 cells) in buffer 
would be very remote. 
The investigators conclude that these 
studies provide overwhelming support 
for the safety of biological containment 
based on poorly mobilizable plasmids 
such as pBR325. 
These results have been presented to 
the Recombinant DNA Advisory 
Committee and have been submitted to 
a scientific journal for publication. 
C. Molecular Basis of E. Coli 
Colonization 
The National Advisory Allergy and 
Infectious Diseases Council had 
supported award of a grant (AI 16370) to 
the University of Rhode Island on behalf 
of Dr. Paul S. Cohen for studies designed 
to elucidate the molecular mechanism of 
intestinal colonization of mice with 
human strains of E. coli. 
At this point in time the majority of 
experiments using recombinant DNA 
technology employ host-vector systems 
based on E. coli K-12 and its plasmids 
or bacteriophages. Prominent umong the 
scenarios raised early in the debate over 
use of this technology was the possible 
colonization of the intestinal tract by 
host-vector systems followed by various 
consequences depending on the 
elaboration of a product which would 
cause harm to the individual by either 
direct or indirect mechanisms. There 
now are a considerable number of 
studies describing the survival of 
various types of E. coli in the intestinal 
tract of man and mice and they 
demonstrate a tremendous disparity in 
the survivability and colonization 
potential of such strains. A complete 
understanding of those factors that 
control survival and colonization may 
permit the development of both safer 
and more useful E. coli hosts in the 
future as well as perhaps provide data 
suggesting adjustments in the physical 
containment requirements of the NIH 
Guidelines governing use of this 
technology. 
Dr. Cohen and collaborators are 
attempting to determine the relative 
importance of plasmid and chromosomal 
determinants of colonization and to 
identify all surface components that are 
associated with this process. The 
investigators have described an animal 
model which they are using to 
systematically investigate the relative 
ability of E. coli strains to colonize the 
large bowel of mice. In this system, the 
colonizing ability of one strain of E. coli 
is determined relative to a second strain 
by simultaneously feeding both strains 
to male CD-I mice whose normal 
facultative flora has been reduced by 
adding streptomycin to their drinking 
water. Subsequent monitoring of the 
level of each strain present in the feces 
provides an estimate of how well one 
strain colonizes the large intestine 
relative to the second competing 
organism. 
Employing this system. Dr. Cohen's 
laboratory has investigated the 
colonizing ability of human fecal 
isolates relative to common laboratory 
struins (i.e., E. coli B and K-12), and 
examined the effect of limited genetic 
alteration on the colonizing abilities of 
isogenic strains of E. coli. The results of 
these studies indicated that laboratory 
strains are poor colonizers (1 x 10* cells 
per gram feces) relative to fecal isolates 
(about 1 X 10* ceils per gram of feces) 
and that limited genetic alterations can 
indeed enhance or reduce the relative 
colonizing abilities of E. coli strains. 
1. Effect of plasmid Cone Expression 
on Colonizing Ability of E. coli IIS in 
Mice. Experiments were performed to 
test whether the colonizing ability of a 
human fecal strain ( E . coli I IS) is altered 
when transformed with plasmids typical 
of those used in recombinant DNA 
experiments. £. coli 1IS. known to be an 
excellent colonizer of humans, was 
transformed with each of three plasmids 
(pBR325, F-amp. and pJBK5) and the 
effect of each plasmid on its colonizing 
ability was determined. Colonizing 
abilities of the plasmid-containing HS 
strains were reduced three orders of 
magnitude relative to the parental HS 
strain. When these strains were cured of 
either pBR325, F-amp, or pJBK5, they 
regained the colonizing ability of the 
original parental strain. The 
investigators conclude that it is clear 
from these data that plasmid gene 
expression can alter the colonizing 
abilities of an E. coli strain. Further 
experiments showed that the colonizing 
abilities of the plasmid-containing 
strains appeared unchanged when these 
strains were fed to mice along with a 
heterologous strain, E. coli F-18. The 
investigators conclude that these data 
suggest that E. coli HS and E. coli F-18 
colonize biochemically distinct sites in 
the mouse intestine. 
2. Relationship Between the Mouse 
Colonizing Ability of a Human Fecal E. 
coli Strain and its Ability To Bind a 
Specific Mouse Colonic Mucous Gel 
Protein. The abilities of three strains of 
E. coli to bind CD-I mouse colonic 
mucous gel were studied. The strains 
employed were E. Coli F-18, an 
excellent mouse colonizer which was 
originally insolated from the feces of a 
healthy human; E. coli F-18 col', a poor 
mouse colonizer derived from F-18 and 
lacking a large plasmid; and E. coli J 5-3, 
a typical K-12 strain, also a poor 
colonizer relative to F-18. When mouse 
colonic mucous gel was isolated and 
methylated with ( 3 H) formaldehyde, it 
was found to consist of approximately 
18 proteins and contained 
approximately 5% hexose by weight. 
Each of the three strains of £. coli were 
able to use the mucous gel as a sole 
source of carbon and nitrogen for 
growth. When the ability of the three 
strains of E. coli to bind mucous gel was 
examined, it was found that F-18 bound 
