! caping phage in nature could further be 
| blocked by adding various conditional 
j mutations which would permit growth 
| only under special laboratory conditions 
I or in a special permissive laboratory host 
|J with suppressor or pro-type (mop, dnaB, 
|| rpoB ) mutations. An additional safety 
I feature would be the use of an r _ nr 
I (hsdS) laboratory host, which produces 
j phage with unmodified DNA which 
should be restricted in r + m + bacteria that 
j are probably prevalent in nature. The 
likelihood of recombination between the 
X vector and lambdoid prophages which 
I are present in some E. coli strains might 
be reduced by elimination of the Red 
: function and the presence of the recom- 
bination-reducing Gam function to- 
gether with mutations contributing to 
the high lethality of the x phage. How- 
ever, these second-order precautions 
might not be relevant if the stability and 
infectivity of the escaping X particles are 
! reduced by special mutations or by pro- 
pagating the highly unstable heads. 
Despite multiple mutations in the 
j phage vectors and laboratory hosts, the 
yield of phage particles under suitable 
laboratory conditions should be high 
(10 10 -10 n particles /ml ) . This permits 
phage propagation in relatively small 
volumes and constitutes an additional 
safety feature. 
The phenotypes and genetic stabilities 
l of the mutations and chromosome alter- 
\ ations included in these x-host systems 
; indicate that containment well in excess 
of the required 10" 8 or lower survival fre- 
| quency for the X vector with or without a 
cloned DNA fragment should be attained. 
Obviously the presence of all mutations 
contributing to this high degree of bi- 
ological containment must be verified 
periodically by appropriate tests. Labora- 
tory tests should be performed with the 
bacterial host to measure all possible 
I routes of escape such as the frequency of 
j lysogen formation, the frequency of 
| plasmid formation and the survival of 
the lysogen or carrier bacterium. Sim- 
ilarly, the potential for perpetuation of 
a cloned DNA fragment carried by in- 
fectious phage particles can be tested by 
challenging typical wild-type E. coli 
strains or a x-sensitive nonpermissive 
laboratory K-12 strain, especially one 
lysogenic for a lambdoid phage. 
In view of the fact that accurate as- 
sessment of the probabilities for escape 
of infections x-grown on r~ m~ Su+ hosts 
is dependent upon the frequencies of r~, 
Su + , and X-sensitive strains in nature, 
investigators need to screen E. coli 
strains for these properties. These data 
will also be useful in predicting fre- 
quencies of successful escape of plasmid 
cloning vectors harbored in r nr Su + 
strains. 
When any investigator has obtained 
data on the level of containment pro- 
vided by a proposed EK2 system, these 
should be reported as rapidly as possible 
to permit general awareness and evalua- 
tion of the safety features of the new 
system. Investigators are also encouraged 
to make such new safer cloning systems 
, generally available to other scientists. 
NIH will take appropriate steps to aid 
See footnotes at end of article. 
NOTICES 
in the distribution of these safer vectors 
and hosts. 
EK3 host-vectors. These are EK2 sys- 
tems for which the specified containment 
shown by laboratory tests has been inde- 
pendently confirmed by appropriate tests 
in animals, including humans or pri- 
mates, and in other relevant environ- 
ments in order to provide additional 
data to validate the levels of contain- 
ment afforded by the EK2 host-vector 
systems. Evaluation of the effects of in- 
dividual or combinations of mutations 
contributing to the biological contain- 
ment should be performed as a means to 
confirm the degree of safety provided 
and to further advance the technology 
of developing even safer vectors and 
hosts. For the time being, no host-vector 
system will be considered to be a bona 
fide EK3 host-vector system, until it is 
so certified by the NIH Recombinant 
DNA Molecule Program Advisory Com- 
mittee. 
2. Classification of experiments using 
the E. coli K-12 containment systems. In 
the following classification of contain- 
ment criteria for different kinds of re- 
combinant DNAs, the stated levels of 
physical and biological containment are 
minimums. Higher levels of biological 
containment (EK3 > EK2 > EK1) are 
to be used if they are available and are 
equally appropriate for the purposes of 
the experiment. 
(a) Shotgun Experiments. These ex- 
periments involve the production of re- 
combinant DNAs between the vector and 
the total DNA or (preferably) any par- 
tially purified fraction thereof from the 
specified cellular source. 
(i) Eukaryotic DNA recombinants — 
Primates. P3 physical containment + an 
EK3 host-vector, or P4 physical contain- 
ment + an EK2 host-vector, except for 
DNA from uncontaminated embryonic 
tissue or primary tissue cultures there- 
from, and germ-line cells for which P3 
physical containment + an EK2 host- 
vector can be used- The basis for the 
lower estimated hazard in the case of 
DNA from the latter tissues (if freed of 
adult tissue) is their relative freedom 
from horizontally acquired adventitious 
viruses. 
Other mammals. P3 physical contain- 
ment + an EK2 host-vector. 
Birds. P3 physical containment + an 
EK2 host-vector. 
Cold-blooded vertebrates. P2 physical 
containment 4- an EK2 host-vector ex- 
cept for embryonic or germ-line DNA 
which require P2 physical containment 
+ an EK1 host-vector. If the eukaryote 
is known to produce a potent toxin, the 
containment shall be increased to P3 + 
EK2: 
Other cold-blooded animals and lower 
eukaryotes. This large class of eukaryotes 
is divided into the following two groups : 
(1 ) Species that are known to produce 
a potent toxin or are known pathogens 
(i.e., an agent listed in Class 2 of ref. 5 or 
a plant pathogen) or are known to carry 
such pathogenic agents must use P3 
physical containment + an EK2 host- 
vector. Any species that has a demon- 
strated capacity for carrying particular 
27917 
pathogenic agents is included in this 
group unless it has been shown that 
those organisms used as the source of 
DNA do not contain these agents; in this 
case they may be placed in the second 
group. 
(2) The remainder of the species in 
this class can use P2 + EK1. However, 
any insect in this group should have been 
grown under laboratory conditions for at 
least 10 generations prior to its use as a 
source of DNA. 
Plants. P2 physical containment + an 
EK1 host-vector. If the plant carries a 
known pathogenic agent or makes a 
product known to be dangerous to any 
species, the containment must be raised 
to P3 physical containment + an EK2 
host-vector. 
(ii) Prokaryotic DNA recombinants — 
Prokaryotes that exchange genetic in- 
formation with E. coli.’ The level of 
physical containment is directly deter- 
mined by the rule of the most dangerous 
component (see introduction to Section 
in) . Thus PI conditions can be used for 
DNAs from those bacteria in Class 1 of 
ref. 5 (“Agents of no or minimal hazard 
***.”) which naturally exchange genes 
with E. coli; and P2 conditions should 
be used for such bacteria if they fall in 
Class 2 of ref. 5 (“Agents of ordinary po- 
tential hazard * * *.”), or are plant 
pathogens or symbionts. EK1 host-vec- 
tofs can be used for all experiments re- 
quiring only PI physical containment; 
in fact, experiments in this category can 
be performed with E. coli K-12 vectors 
exhibiting a lesser containment (e.g., 
conjugative plasmids) than EK1 vectors. 
Experiments with DNA from species re- 
quiring P2 physical containment which 
are of low pathogenicity (for example, 
enteropathogenic Escherichia coli. Sal- 
monella typhimurium, and Klebsiella 
pneumoniae ) can use EK1 host-vectors, 
but those of moderate pathogenicity (for 
example. Salmonella typhi, Shigella dys- 
enteriae type I, and Vibrio cholerae) 
must use EK2 host-vectors.* A specific 
example of an experiment with a plant 
pathogen requiring P2 physical contain- 
ment + an EK2 host-vector would be 
cloning the tumor gene of Agrobacterium 
tumefaciens. 
Prokaryotes that do not exchange ge- 
netic information with E. coli. The mini- 
mum containment conditions for this 
class consist of P2 physical containment 
+ an EK2 host-vector or P3 physical 
containment an EK1 host-vector, and 
apply when the risk that the recombi- 
nant DNAs will increase the pathogenic- 
ity or ecological potential of the host is 
judged to be minimal. Experiments with 
DNAs from pathogenic species (Class 2 
ref. 5 plus plant pathogens) must use 
P3 + EK2. 
(iii) Characterized clones of DNA 
recombinants derived from shotgun ex- 
periments. When a cloned DNA recom- 
binant has been rigorously character- 
ized 4 and there is sufficient evidence that 
it is free of harmful genes, 4 then experi- 
ments involving this recombinant DNA 
can be carried out under PI + EK1 con- 
ditions if the inserted DNA is from a 
FEDERAL REGISTER, VOL. 41, NO. 131 — WEDNESDAY, JULY 7, 1976 
