Figure 1

Figure 1. Comparison of the van gene cluster from the glycopeptide producer Streptomyces toyocaensis (32) and of the vanA operon (33) from gram-positive cocci. Open arrows represent coding sequences...

One should distinguish "natural" antimicrobial drugs (e.g., kanamycin), which are produced by microorganisms from the environment, from semisynthetic (e.g., amikacin) and entirely synthetic compounds (e.g., quinolones), which are produced, at least in part, by humans. The microorganisms that produce natural antimicrobial drugs have to protect themselves from the products of their own secondary metabolism. To ensure their survival, these organisms have developed self-protection mechanisms similar to those found in resistant human pathogens (7); this observation led to the idea that the producers constitute the pool of origin of certain resistance genes (8). Therefore, the study of resistance in the strain used for the industrial production of an antimicrobial agent could allow a strong prediction about the mechanism that will be found later in bacteria pathogenic for humans. For example, the study of glycopeptide producers would have allowed the elucidation, long before it actually occurred, of the mechanism by which enterococci and, more recently, staphylococci could become resistant to these drugs (Figure 1).

As already noted, bacteria are resistant to antimicrobial drugs after horizontal DNA transfer or mutations. Thus, another prediction that can be made is that bacteria will transfer to susceptible species, resistance determinants already known in other bacterial genera, for example, the recent acquisition of glycopeptide resistance by Staphylococcus aureus from Enterococcus spp (9). However, this prediction is limited since it refers to mechanisms that have already been explained. In addition to being antimicrobial agent producers, the commensal bacteria of mammals, particularly those in the gut, also represent a pool of origin for resistance genes. When infections are treated with an antimicrobial agent, all bacteria in the host are affected, including the commensal flora, which could result in the selection of resistant commensals, particularly in children who are administered oral antimicrobial drugs too frequently. Large numbers of these resident bacteria are present in the digestive tract where they are often in transient, but intimate, contact with exogenous microorganisms that are in various developmental states, including competence. These conditions favor the transfer of genes by transformation and by conjugation. Including antimicrobial drugs in animal feed also leads to the selection of a pool of resistance genes that can be transferred to commensal bacteria in the human digestive tract and thus ultimately to human pathogens, even when selective pressure is absent (10).

In the case of mutations, predictions can be supported by 2 types of experimental approaches: in vivo with intact bacteria or in vitro by using DNA. Mutations resulting in resistance can be obtained in an accelerated fashion by using hypermutators, that is, bacteria deficient in the DNA repair system (11). Mutations are also accumulated by using continuous cultures, preferably in chemostats under suitable selective pressure. A similar enhanced rate of evolution can be obtained by (saturated) DNA mutagenesis, followed by transformation into an appropriate host. This technique, for example, was used successfully to study the extent of variations in penicillinase genes that generate extended-spectrum β-lactamase agents (12).

Genes from gram-positive cocci can be transferred by conjugation (of plasmids or transposons) not only among these microorganisms but also to gram-negative bacteria (30). The reverse is not true because of limitations in heterologous gene expression. Consequently, one can confidently predict further dissemination of the resistance gene pool of gram-positive to gram-negative bacteria.

We have been aware for a long period that "everything that exists in the universe is the result of chance and necessity" (Democritus, 460–370 BC), which holds true for antimicrobial drug resistance. Most unfortunately, and for various reasons, it is extremely difficult to think like a bacterium. In other words, predicting the emergence of resistance to a drug class by a precise molecular mechanism is nearly impossible (e.g., glycopeptide resistance in enterococci or plasmid-mediated resistance to fluoroquinolones). We also cannot anticipate, among all the conceivable mechanisms of resistance (31), which will emerge first under natural conditions. However, based on the understanding during recent decades of the physiology (genetics and biochemistry) of bacterial resistance to antimicrobial drugs, impressive progress has been made in the techniques for in vitro detection and for elucidation of resistance. This progress should, in turn, be helpful in delaying the second step of resistance: dissemination.