83 
back and do other experiments. We have, in some cases, changed the require- 
ments for using the EK2 systems to meet what we thought were tighter re- 
quirements, and the result is that I think we have a clearer understanding 
of what we can expect as EK2 systems. I would like you to keep that in mind, 
particularly when we get to the next section when we talk about particular 
guidelines for containment, that these EK2 systems are an essential part of 
that containment. 
Finally, we get to HV3, which, once again, is the only real candidate 
in the near future, is an E^ col i K-12 system. The requirements for HV3 
have been changed somewhat from the original Guidelines, and the proposed 
revision tightens the requirements for HV3 over those in the original Guide- 
lines. The new requirements are in response to comments that we have received 
throughout the last year and a half, and faults that some of us saw in the 
original EK3 definition. It retains the original requirements for EK3 that 
both j_n vivo and j_n vitro testing be done on any proposed EK3 system, and 
that that be done by some independent body other than the developers of the 
system. 
The new requirements are, one, that the vector be extremely defective in 
mobi 1 izabi 1 i ty , or if at all possible, host-dependent. That is, that the 
vector be essentially non-transmi ssible from the host in which it is certi- 
fied, either by mobilization or by any other mechanism. We have put this 
in. Although that was certainly something that was wished in the original 
Guidelines, it wasn't stated nearly as explicitly as it is in the proposed 
revision. The aim of this restatement is to promote a particularly careful 
look at this data by any committee considering EK3 host-vectors. 
Finally, the other major change in HV3 is a new requirement that no 
clinically or agriculturally important antibiotic-resistance marker be 
functional on a vector to be used for recombinant DNA research. The reason 
for this — I think it has already come up a little bit earlier this morning — 
is to prevent any inadvertent advantage for a host carrying a clone which 
also, for instance, has ampicillin resistance and encounters ampicillin 
someplace. It has managed to escape our physical containment and it hasn't 
died as it was supposed to and now it encounters ampicillin and has such a 
great advantage over everything else around it that it is enriched for. 
Although this may be an unlikely scenario, we thought that it was possible 
to block this by putting this requirement in for HV3, and therefore we have 
introduced it. So I think now we can say with confidence that an HV3 system 
which meets all of these criteria is just about as safe as we think we can 
make it. 
The other major changes in this section of the Guidelines are not 
really changes in practice. They are more inclusion in the Guidelines of 
practices that have already been used in the last year by the Recombinant 
Advisory Committee. They just make those practices explicit, and hopefully 
a little bit clearer to the world at large. For example, the kinds of 
things that have been put in, the use of expert working groups to consider 
[ 287 ] 
