33 
In general, the strain of any prokaryotic species used as the 
host should conform to the definition of Class 1 etiologic agents 
given in ref. 5 (i.e., "Agents of no or minimal hazard "), and 
the plasmid or phage vector should not make the host more hazardous. 
In addition, it is recommended that the newly developed host-vector 
systems offer some distinct advantage over the E. col i K-12 host- 
vectors--for instance, thermophilic organisms or other host-vectors 
whose major habitats do not include humans and/or economically im- 
portant animals and plants. Appendix A gives a detailed discussion 
of the B_. subti 1 is system, the most promising alternative to date. 
At the initial stage, the host-vector should exhibit at least 
a moderate level of biological containment comparable to EK1 systems, 
and be capable of modification to obtain high levels of containment 
comparable to EK2 and EK3. The type of confirmation test(s) required 
to move a host-vector from an EK2-type classification to an EK3-type 
will clearly depend upon the preponderant habitat of the host-vector. 
For example, if the unmodified host-vector propagates mostly in, on, 
or around higher plants, but not appreciably in warm-blooded animals, 
modification should be designed to reduce the probability that the 
host-vector can escape to and propagate in, on, or around such plants, 
or transmit recombinant DNA to other bacterial hosts that are able to 
occupy these ecological niches, and it is these lower probabilities 
which should be confirmed. The following principles should be followed 
in using the containment criteria given for experiments with E_. col i 
K-12 host-vectors as a guide for other prokaryotic systems. Experi- 
ments with DNA from prokaryotes (and their plasmids or viruses) 
should be classified according to whether the prokaryote in question 
[ 104 ] 
