APPENDIX O 
"virulence factor" is usually applied to only those characteristics of 
pathogenic bacteria which distinguish them from the common and harmless 
non-pathogenic microorganisms. Nevertheless, it is obvious that the num- 
ber of characteristics which a bacterium must have in order to cause 
disease is rather enormous and is by no means confined to those traits 
commonly listed under the heading of "virulence factors". This consi- 
deration is important in the evaluation of the so-called EK-2 and EK-3 
("enfeebled" or "disarmed") strains of coli which are to be used in 
most recombinant DNA research. These microorganisms have been deliberately 
deprived of a number of those metabolic or structural features which are 
required for the normal functions of any common bacterium. For this reason 
EK-2 and EK-3 strains of _E. coli can grow only under those special labor- 
atory conditions which provide the bacterial equivalent of our lawyer's 
heart-lung machine, artificial kidney, intravenous feeding device, etc. 
Thus, each deletion of an ordinary metabolic or structural trait from an 
"enfeebled" E. coli strain is at the same time also a deletion of an 
indispensible link in the long chain of virulence factors which this strain 
would have to possess if it were to become a serious pathogen. 
As the reader will recall, our original argument states that coli 
is a common and usually harmless bacterium which is not a highly dangerous 
pathogen because it lacks a large number of virulence factors and that, 
consequently, the inadvertent introduction by recombinant DNA methods of 
genes coding for only one or a few of these factors would not be sufficient 
to restore the entire complement of traits necessary for virulence. But, 
one may ask, how can we be sure that coll reallylacks a large number 
of virulence factors? Is it not possible that coll may in fact possess 
all but one of the factors required to turn it into a pathogen capable of 
causing serious epidemics among human populations? This is a reasonable 
question and there are indeed precedents. For example, certain bacteria 
which often inhabit the normal human throat are close relatives of the 
diphtheria bacillus and some of these, as far as we know, differ from this 
pathogen , only in that they lack the gene for toxin production. Moreover, 
when 2 - week old chicks were injected intravenously with a rather enormous 
quantity of JE. coll (approximately a billion cells, enough to overcome most 
body defense mechanisms by sheer bacterial numbers), the introduction of a 
limited amount of genetic information allowed these bacteria to kill the 
animals more efficiently (4). It is therefore possible, that the intro- 
duction of a limited amount of genetic material into coli may augment 
those disease-causing properties which these bacteria already possess . 
It is also true, that some strains of coll cause diarrheal disease in 
man, others are common causes of urinary tract infections. In view of 
this I would consider it still highly unlikely, but certainly not entirely 
impossible that the introduction of a single gene into an coll strain 
might convert it into a more efficient causative agent of infectious 
diarrheal or urinary tract disease. However, urinary tract infections 
by JE. coll are not contagious at all and enteric infections by special 
strains of coli are self-limiting and not life threatening (except 
perhaps in newborns) and do not give rise to widespread epidemics in 
countries with adequate sanitation, because the route of dissemination 
Appendix 0--3 
