Antibiotic inactivation mechanisms share many similarities with well-characterized enzymatic reactions. Hydrolysis, group transfer, and redox enzymes are all involved in primary and intermediary microbial metabolism and, thus, likely serve as the origins of resistance. …primary sequence analysis of resistance proteins, and in particular determination of their molecular mechanisms and three-dimensional structures, has revealed homologies to known metabolic and signaling enzymes with no antibiotic resistance activity. Therefore, one can speculate that these are the original sources of resistance. A reasonable scenario, then, would have such housekeeping genes encoding enzymes with modest and fortuitous resistance properties evolving as a result of exposure to the antibiotic into a bona fide and efficient resistance enzymes…. The antibiotic-producing organisms are possible originators for many resistance enzymes, as here resistance and biosynthesis must co-evolve and, thus, the evolutionary pressure would be chronic. Furthermore, the soil environment where many of these organisms reside is home to neighboring species that produce their own antibiotics, thereby adding to the evolutionary pressure to develop resistance enzymes. — Gerard D. Write (2005) (p. 1464)
Chromosomal resistance is generated by mutations and extrachromosomal genes can be modified by mutations. In either case, what is occurring with mutation is the modification of the nucleotide sequence of a gene, thereby creating a new and perhaps novel allele at an existing locus. Mutation is the ultimate source of genetic variation, though recombination also can give rise to modification of the nucleotide sequence of a gene, again creating new and perhaps novel alleles. The difference is that recombination does not involve changes to what nucleotides are present but instead changes to the order of the sequence of existing nucleotides. In either case, it is not that new loci are formed within a genome but instead what are formed are variations on loci that already exist, that is, the creation of new alleles.
The modification of an existing locus frequently results in a loss of function. Even when there is a gain in function there may be a corresponding loss of function, so-called antagonist pleiotropy where enhanced functionality of an allele in one circumstance gives rise to reduced functionality in a different circumstance. Indeed, in a sense these alleles can possess a pleiotropy consisting of, on the one hand, their normal function and on the other hand antibiotic resistance. Mutations, even when they are helpful, thus can be costly. This certainly can be true for chromosomal resistance since it is an existing function or structure that is being modified by the mutation, such as of a ribosome. The evolution of new abilities, in other words, typically involves tradeoffs of one sort or another, and evolution of antibiotic resistance through the mutation of existing alleles certainly is no exception.
By contrast, extrachromosomal resistance may be more robust to mutational change. This reduced likelihood of pleiotropic costs is not a function of the location of these loci but instead of the nature of their utility. In particular, if a gene has essentially a single utility, and that utility is only transiently present, then the result of mutation is likely to be reduction in function, gain in function, or no change in function, that is, rather than gain in function in one circumstance but loss of function in another. Thus, generally, mutations impacting extra-chromosomal antibiotic resistance genes – or indeed redundant genes generally – in principle may be less likely to be detrimental to the organism as a whole, and this can be so even if the gene is currently useful simply because the gene in fact possesses only a single use, that of effecting resistance. Furthermore, to the extent that the gene only minimally interacts with other genes within the bacterium, or indeed other genes are not dependent on its structure or functioning, which is also more likely given the nature of extrachromosomal genes as accessory or transient within genomes, then the potential for mutations to be costly may be further reduced relative to the costs associated with mutations to, for example, constitutively expressed housekeeping genes.
The result may be an enhanced ability to explore sequence space, that is, faster adaptive evolution among these accessory genes than among non-accessory genes. Furthermore, as accessory genes on plasmids often are found in multiple copy number (that is, multiple-copy plasmids) and are often not relevant to bacterial survival (such as when antibiotics are not present in the environment) then resistance genes may be viewed in some ways as being able to evolve in ways that are equivalent to the evolution of pseudogenes, with occasional alleles that display increased benefits to the carrying bacterium increasing their representation within a population due to natural selection acting on the population. On the other hand, the presence of multiple plasmid copies likely means that luck is involved in whether a beneficial extrachromosomal allele becomes fixed even within the bacterium within which it is present. Nonetheless, between a potential for these genes to be transmitted to new hosts and also to be mutated with less consequential cost, extrachromosomal resistance genes, or indeed extrachromosomal genes in general may be particularly capable of evolutionary experimentation towards the development of new functions during the course of bacterial evolution in nature.
Indirect evidence suggests that selective pressures have mobilized ARGs [Antibiotic Resistance Genes] from their initial chromosomal location in bacteria . Phages persist better in aquatic environments than their bacterial hosts and, due to their structural characteristics, better than free DNA . This higher survival and the abundance of phages carrying ARGs in animal and human wastewater support the notion that phages are vehicles for mobilization of the environmental pool of ARGs that contribute to the maintenance and emergence of new resistances. — Gerard D. Write (2005) (p. 1464)
The movement of resistance alleles between bacteria can occur in association with mobile genetic elements (via conjugation or specialized transduction) or alternatively by more random mechanisms of mobility (generalized transduction, re: the quote immediately above, and transformation). More randomized movement can be rare, perhaps more rare than simply mutation to antibiotic resistance. In addition, such movement is poorly documented, except for the occasional movement of plasmids between bacteria due to generalized transduction. In this section I therefore briefly concentrate on the issue of conjugative movement of resistance genes, that is, movement on plasmids, versus the movement of resistance genes as loci carried by phages, particularly as so-called phage moron genes. In all cases these represent mechanisms of horizontal gene transfer, and therefore sex, as experienced by bacteria.
Recall that phages, or bacteriophages, are viruses that infect bacteria. Morons are pieces of DNA that constitute "more" or at least different DNA from that normally found in association with a phage's genome. This DNA presumably is acquired by illegitimate recombination processes, and indeed the genes found on plasmids should follow similar acquisition pathways. Though resistance genes are a fairly normal aspect of plasmid biology, this is much less the case for phages. Why should that be? We can consider two possible explanations.
The first is that phages are much more aggressive in their acquisition of new hosts than are plasmids. This means that there may be less of a need for phages to carry genes that are potentially beneficial to their hosts, though certainly they do carry such genes, re: lysogenic conversion. Another way of saying this is that at steady state the rate of phage acquisition of bacteria likely is higher than that of plasmid acquisition of bacteria so therefore a higher rate of phage failure to become established within a given bacterial lineage may be tolerated without overall extinction of the phage population. Phages therefore do not necessarily need to place as many bets to obtain a payoff, that is, do not need to carry as many genes that are helpful to potential bacterial hosts in comparison to plasmids because individual attempts to establish themselves within a bacterial lineage may be much less dear for phages than for plasmids. In addition, and consistently, a large fraction of phages are not temperate so do not establish long-term, potentially mutualistic relationships with their hosts at all, yet these phages still survive. Even among temperate phages, that is, those capable of displaying lysogenic infections, a large fraction of phage infections of individual bacteria do not even attempt to establish such mutualistic interactions.
By way of analogy, if there were only one store in town, you might be tempted to put forth a great deal more effort towards keeping that store owner on your good side than if instead there were 100 or 1,000 stores to choose from. Interestingly, similar arguments can be made in terms of the evolution of pathogen virulence where the "niceness" displayed by symbionts towards host organisms tends to be greater the less likely transmission to new host organisms may be. Plasmids thus may need to display a greater variety of tendencies towards "niceness" than do phages, including in terms of multiple resistance genes, simply because the plasmid potential to transmit themselves to new hosts may be more limited than the phage potential.
The second issue is one of limitations on gene acquisition where phages are limited by genome packaging constraints that presumably do not apply as readily to plasmids. In other words, phages may not be able to encode multiple resistance genes even if they needed to, or at least would have to crowd those genes into their genome at the expense of other genes that might be useful to the phage. In addition, for temperate phages the lysogenic state, which is when these genes would be useful, represents only one aspect of their replication strategies and enhancement of the lysogenic state could, in terms of antagonistic pleiotropies, result in reduced ability to produce and disseminate phage progeny. Plasmids likely also display tradeoffs between horizontal gene transfer to new hosts and benefitting current hosts, though I speculate that these constraints are greater for temperate phages than they are conjugative plasmids. Notwithstanding these issues, phages do carry genes that are beneficial to their hosts, giving rise to what is known as lysogenic conversion. Only relatively rarely, however, do these genes tend to be ones that provide antibiotic resistance to bacterial hosts (Galán, 2008) .
Genetic drift should play a role in the evolution of resistance particularly at times when such resistance is not required by an organism, such as during periods when for example antibiotics are not present in a bacterium's environment. As noted, to a degree the loss of an entire plasmid during at least partial down times may be avoided, in part, through the encoding of multiple resistance mechanism. In addition, there are other sorts of not-always-required genes as well as infectious plasmid transmission to new hosts that can be aids to plasmid persistence within populations or communities of bacteria. There also are various mechanisms that plasmids employ to resist being lost from carrying bacteria (i.e., so-called plasmid addiction systems). The encoding of multiple beneficial genes also helps to address the issue of plasmid usefulness when first introduced into individual bacteria and their populations, a time when genetic drift may particularly impact plasmid survival since initial numbers are small.
These mechanisms do not address the question of how individual genes avoid being lost from plasmid populations, however, particularly if the utility of those individual genes is infrequent. Selection, though, might counter drift in these instances if compensatory mutations have become fixed that give rise to reduced bacterial fitness given loss of resistance alleles, even under circumstances where resistance itself is no longer required. That is, and as noted, compensating alleles too can display what in a sense is an antagonistic pleiotropy with different degrees of contribution to fitness depending upon the encoding organism's genetic background.