3S162 
NOTICES 
Cion studies with primary tumors and 
transformed tissue culture cells indi- 
cate that large numbers of cells are re- 
quired to initiate tumor growth; in 
general, only after adaptation effected 
by long passage in animals do small 
numbers of cells produce tumors (Sjo- 
gren. 1964; Klein, 1975). In this regard, 
it should be noted that there Is over- 
whelming evidence that humans are 
highly effective In averting tumorigen- 
esis by DNA viruses. Man. and many 
other vertebrates as well, are infected 
repeatedly with DNA viruses (e.g.. pa- 
povaviruses and adenoviruses) that 
contain the transforming regions dis- 
cussed above (Huebner et al.. 1954; 
Hllleman. 1957; Shah et al.. 1973; Pad- 
gett and Walker. 1973; Brown et al.. 
1975; Shah and Nathanson. 1976). 
Some of these viruses are even known 
to Induce tumors in laboratory ro- 
dents. yet. despite intensive search, 
none of these has been reproduclbly 
associated with the etiology of any 
malignancy In humans (Mackey et al.. 
1976; Piori and DiMayorca. 1976; Wold 
et al.. 1978; Israel et al.. 1978b >. 
The risk scenario whereby a cDNA 
transcript of a positive strand RNA 
virus leads to Infection of the host was 
considered to have a finite but low 
probability. In this model, extralntes- 
tinal Infection by the bacterium carry- 
ing the recombinant DNA results In 
the recombinant DNA molecule, or 
plus strand RNA transcripts thereof, 
entering Into host cells with resultant 
productive viral Infection. Neither car- 
ries a risk if the cDNA Insert Is not a 
complete copy of the RNA genome. 
Unless a deliberate attempt Is made to 
select full-length cDNA molecules for 
ligation to prokaryotic DNA vectors, 
the vast majority of recombinant DNA 
molecules will contain only a segment 
of viral specific DNA and. as a conse- 
quence. the likelihood of cloning the 
whole genome will be low. 
The major block to synthesis of in- 
fectious RNA from the recombinant 
DNA molecule either in the prokaryo- 
tic or eukaryotic cell would be the in- 
accurate Initiation of transcription. 
There is no reason to believe that the 
5 terminus of the viral genome would 
generate a binding site for either pro- 
karyotic or eukaryotic RNA polymer- 
ases following reverse transcription 
into DNA. Rather. RNA synthesis 
would be Initiated either. 
(1) Internally, by a random Initi- 
ation process which would yield nonin- 
fectious molecules; 
(3) Upstream, either at a prokaryotic 
promoter or at a random initiation 
site, in the eukaryotic cell which 
would generate a leader sequence lack- 
ing the appropriate recognition signals 
for binding to eukaryotic ribosomes 
(Shine and Dalgamo. 1974; Steitz and 
Jakes. 1975; Hagenbuchle et al.. 1978). 
To produce Infectious RNA from such 
a primary transcript, a site-specific 
cleavage would have to occur; 
(3) At the 5 terminus by random ini- 
tiation. but this undoubtedly would be 
an extremely rare event. 
Any full-length plus strand transcripts 
synthesised in E. coli would have to 
escape degradation by extracellular ri- 
bonucleases following their release 
from bacteria and then Initiate an in- 
fection in a sensitive cell, which, under 
the best conditions in the laboratory, 
is an extremely inefficient process 
(Ellem and Colter. 1961). 
Further studies of this type of viral 
DNA Insert would clearly be desirable. 
The cloning of the complete genome 
of DNA viruses (Including proviral 
forms of retroviruses) prorides a sce- 
nario felt to carry a finite probability 
of risk, but this too is considered ex- 
tremely low. Since viral DNA mole- 
cules are generally infectious, any situ- 
ation in which the complete viral 
genome, without deletion or substitu- 
tion. is excised from the recombinant 
molecule could conceivably lead to in- 
fection. In particular, circular DNA 
genomes, cleaved with a single cut re- 
striction enzyme and ligated to a pro- 
karyotic vector, provide such a condi- 
tion; however, for accurate excision 
one must postulate the unlikely even- 
tuality of the recombinant molecule 
being brought in contact with the 
same restriction enayme used during 
the Insertion of the DNA segment into 
the vector. Complete viral genomes 
could also be generated from a recom- 
binant molecule in the case of oli- 
gomers of viral DNA inserts; a com- 
plete copy of the viral DNA could then 
be generated by Intramolecular recom- 
bination. Full length linear DNA mole- 
cules inserted into a vector by oligo 
dA-oligo <JT tailing or by addition of 
other linker molecules would be un- 
likely to excise accurately terminating 
genomes. 
Given that it is conceptually possible 
to generate Infectious molecules from 
recombinant DNA containing the com- 
plete genome of DNA viruses, it is im- 
portant to assess the factors that 
would result in such Infectious DNA 
causing disease. The probability that 
the excised molecules would infect 
host cells is of course relatively low in 
view of the inefficiency of infection by 
naked DNA (Ito. 1960: Burnett and 
Harrington. 1968; McCutchan and 
Pagano. 1968; Mayne et al.. 1971; 
Graham and van der Eb. 1973: Sol and 
van der Noordaa, 1977). Another im- 
portant factor in such a scenario is the 
susceptibility of neighboring cells to 
the virions produced by a cell infected 
with the infectious DNA. If the sur- 
rounding cells were not sensitive to 
the virus itself, the initially infected 
cell, producing virions, could not 
spread this infection to other cells or 
tissues: in the absence of virus spread 
there would be no risk other than the 
possible transformation of a small 
number of cells, in the case of onco- 
genic viruses, as discussed above. Only 
if the host was sensitive to the virus 
would spread occur; this would not be 
different from infection with the virus 
itself and would not generally repre- 
sent a unique biohazard. Futher study 
of this type of viral DNA insert would 
be extremely useful. 
Recapitulation. It should be noted 
that these model scenarios were not 
constructed from an anthropocentric 
viewpoint, but apply generally to any 
model in which vertebrates are colo- 
nized by the E. coli carrying the re- 
combinant molecule. 
To recapitulate, even assuming a 
worst case situation, this analysis 
leads us to conclude that cloning the 
subgenomic segments of nononcogenic 
viruses, the complete genome of nega- 
tive strand RNA viruses, and any part 
of the genome of segmented viruses 
carries no risk of generating a bioha- 
zard. Second; It is possible to construct 
conceivable but extremely unlikely 
sets of circumstances resulting in a 
biohazard from the cloning of the 
transforming segment of oncogenic 
viruses. And. third: It is possible to en- 
visage feasible biohazard scenarios 
from cloning of the complete genome 
of DNA viruses or the entire genome 
of plus strand RNA viruses but even 
these carry little possibility of risk. 
When this analysis is combined with 
the Immense unlikelihood of generat- 
ing this worst case scenario in the first 
place, given good laboratory practice 
and the safety Inherent in use of ap- 
proved E. coli K-12 host-vector sys- 
tems (Gorbach. 1978), the group felt 
strongly justified in concluding from 
available scientific information that 
viral genomes or fragments thereof, 
cloned in E. coli K-12 using approved 
plasmid or phage vectors pose no more 
risk than work with the Infectious 
virus or its nucleic acid and in most, if 
not all cases, clearly present less risk. 
In fact, the Workshop participants 
agreed that cloning of viral DNA in E. 
coli K-12 may provide a unique oppor- 
tunity to study with greatly reduced 
risks the biology of extremely patho- 
genic and virulent viruses. 
The group also agreed that the clon- 
ing of cDNA copies of viroids in E. coli 
K-12 should be postponed until more 
information is available about their 
molecular and cellular biology. 
Recommendation. Based on these 
considerations the participants of the 
U.S.-EMBO Workshop concluded that 
the use of P2 (NIH guidelines) or Cl 
(Williams report) containment meas- 
ures. in conjunction with an EK1 host- 
vector system should provide adequate 
containment for cloning any viral 
genome or fragment thereof and rec- 
ommended this as the minimum con- 
FEDERAl REGISTER VOL 43, MO. 144— FRIDAY. JULY 28. 1978 
[ 123 ] 
