INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

How much diversity exists in the antibiotic resistance patterns within natural populations of enteric bacteria, and how have the levels of resistance changed since the use of antibiotics became widespread? Increased introduction of antimicrobial agents into the environment via medical therapy, agriculture, and animal husbandry (5, 8) has resulted in new selective pressures on bacterial populations. This has exacerbated the problem of controlling microbes in a disease setting and has caused a resurgence of bacterial diseases worldwide due to the acquisition and transfer of virulence factors and antibiotic resistance genes (17, 30). These problems are further compounded by the persistence of resistance determinants in bacterial genomes over hundreds of generations, even in the absence of antibiotics as selective agents (14).

There are two general categories of antibiotic resistance traits displayed by microorganisms: (i) those that allow microorganisms to withstand relatively high levels of a specific antimicrobial agent, which are conferred by mutations in genes responsible for antibiotic uptake or binding sites, as well as those gained by acquisition of genes on mobile elements (6, 7, 18); and (ii) those provided by genes conferring nonspecific low-level resistance to multiple antibiotics, such as the multiple antibiotic resistance (mar) locus (1, 9, 11). Although dissemination of resistance genes by mobile elements provides a rapid response to antibiotic challenge, increases in background levels of resistance might also be expected in natural populations in environments augmented with antibiotics.

To examine the ways in which increased exposure to antibiotics over the past several decades has changed the patterns and levels of resistance in natural populations of bacteria, we compared MICs for enteric strains isolated before the use of antibiotics became widespread to MICs for contemporary populations that vary in exposure to antibiotics due to their host environments. Among pre-antibiotic era strains the prevalence of genes conferring very high levels of resistance is known to be low due to the lack of strong selective pressures (13, 16). Similarly, the background levels of resistance in these strains might also be expected to be lower than those of strains isolated since the use of antibiotics became widespread even in contemporary strains not collected as a direct result of infection and antibiotic treatment.

Host ecology and environment also factor into the patterns of antibiotic resistance in natural populations. For example, strains of Salmonella are likely to have experienced different selective pressures for resistance than Escherichia coli strains have experienced because they reside primarily in nonmammalian hosts and are far less likely to have encountered most commercial antibiotics in their natural environments. However, such organisms have been exposed to naturally occurring antimicrobial agents, such as the small-polypeptide defensins, that are distributed broadly in mammalian and nonmammalian hosts. Hence, the patterns of resistance of enteric bacteria to these antimicrobial agents likely differ from the patterns produced as a result of challenge by commercially applied antibiotics. By determining the MICs of several antibiotics for bacterial collections representing different temporal and host populations, we determined that the environmental occurrence of antibiotics is reflected in the resistance patterns displayed by enteric bacteria and that the incidence and levels of antibiotic resistance in bacterial strains have changed in response to the increased application of antibiotics.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains. The collections of natural populations of enteric bacteria chosen for this study were as follows. (Full details concerning the sources, strain designations, and years of isolation are given in the references cited.) (i) The ECOR collection (22) of E. coli strains contains 72 strains isolated from natural populations between 1972 and 1982. The strains were derived from numerous host types, including healthy humans from New York, Iowa, and Sweden, humans with urinary tract infections, and domesticated, zoo-kept, and wild-caught mammals. Strains in this collection were originally selected to encompass a broad range of hosts and geographic locations and to represent the genotypic diversity within this species, as determined by multilocus enzyme electrophoresis (12, 25). The MICs of several antibiotics and the presence of integrons near antibiotic resistance cassettes have been surveyed with this collection (18). (ii) The SARC collection of Salmonella strains contains 20 strains isolated between 1958 and 1987 representing the major subspecific groups of Salmonella enterica (groups I, II, IIIa, IIIb, IV, VI, and VII) and Salmonella bongori (group V), as defined by genetic and phenotypic characteristics (3). Whereas members of these species normally circulate in nonmammalian hosts, as represented by strains isolated from a snake, a frog, and a bird, most of the SARC strains were recovered from humans with opportunistic infections. (iii) Bacteria from the pre-antibiotic era were represented by E. coli strains amassed by E.D.G. Murray (13). These strains were recovered primarily from humans with infections between 1885 and 1941 and were collected over a wide geographic range. The antibiotic resistance profiles of these and other natural strains of E. coli have also been reported by Routman et al. (24).

Antibiotic resistance profiles. The MICs of five antibiotics (ampicillin, chloramphenicol, kanamycin, tetracycline, and protamine) were determined by the gradient plate method of Szybalski and Bryson (29). Most of the gradient plates contained 1.5% Luria-Bertani (LB) agar; the exceptions were the plates containing protamine, which contained 1.1% LB agar. When plates were prepared, LB agar containing the appropriate antibiotic was poured as the bottom layer because this method produced more accurate and reproducible results. Plates were incubated at room temperature overnight prior to use to ensure proper diffusion of the antibiotic. Streaked plates were incubated overnight at 37°C, and the MIC for a strain was obtained by averaging the results for two to four replicates and then rounding to the nearest whole number for all of the antibiotics except protamine. Statistical tests were performed in Statview 1.0, and the differences in mean values were compared by Mann-Whitney U tests.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

MIC profiles for strains from each of the three collections are shown in Tables 1 through 3. For each of the commercially administered antibiotics (ampicillin, chloramphenicol, kanamycin, and tetracycline), the distributions of MICs were essentially bimodal, with the two modes corresponding to high-level and background resistance. (The exact distributions were different for different antibiotics, but in general, high-level resistance means resistance to concentrations of >50 µg/ml, whereas background levels were less than 10 to 20 µg/ml depending on the antibiotic.)

High-level resistance to antibiotics. High-level resistance to commercially applied antibiotics was more common among contemporary strains than among strains from the pre-antibiotic era. (In fact, Murray strain 125, the single pre-antibiotic era strain displaying high-level resistance, was probably a contaminant because its MIC profile did not correspond to that reported previously [13].) Approximately 20% of strains in both the ECOR and SARC collections displayed high-level resistance to at least one antibiotic, and less than 5% were resistant to two. The overall frequencies of resistant strains were similar to those reported previously for these and other antibiotics surveyed with the ECOR collection (18). For the majority of antibiotics, the incidences of high-level resistance were similar for strains from human and nonhuman hosts; however, the rare cases of high-level resistance to kanamycin and high-level resistance to chloramphenicol were detected only among strains from nonhuman hosts. Within the ECOR collection, there was a prevalence of high-level resistance to tetracycline (14% of the strains were resistant to concentrations of >50 µg/ml), whereas 20% of the strains in the SARC collection showed high-level resistance to ampicillin.

Analysis of background resistance patterns. The mean MICs of kanamycin, ampicillin, tetracycline, chloramphenicol, and protamine for each of the collections are given in Table 4. Even when the strains displaying high-level resistance to a particular antibiotic were eliminated from the comparisons, the average MIC of each antibiotic was significantly higher for the ECOR collection than for the Murray collection. Similarly, the mean MICs for the ECOR collection were significantly higher than those for the SARC collection for all antibiotics except protamine (to which the salmonellae were significantly more tolerant).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Expanded application of antibiotics has caused an increase in the incidence of resistance to these antimicrobial compounds, even within bacterial species that are not directly subject to antibiotic control. Numerous genes conferring resistance to antibiotics are presently circulating in bacterial populations, and such factors were not as prevalent prior to the selective pressures produced by the increased use of antibiotics (6, 13). Therefore, it is not surprising that contemporary strains of both E. coli and Salmonella frequently display resistance to high levels of commercially administered antibiotics, and such cases have been documented repeatedly with these and other bacteria (10).

However, aside from the anticipated increase in the incidence of high-level antibiotic resistance in contemporary populations, we also detected a change in the background levels of resistance. Even when all cases of high-level resistance were excluded from our analyses, the recently isolated strains of E. coli displayed significantly higher MICs than the pre-antibiotic era Murray strains displayed. This shift was not an artifact of sampling two genetically distinct or restricted subsets of the E. coli population; strains from the Murray collection represent four of the five phylogenetic groups of E. coli and encompass almost the same range of genotypic variation and the same levels of genic diversity (H) as the ECOR collection strains encompass (H_ECOR is 0.34 and H_Murray is 0.27 based on electrophoretic variation at 35 and 10 enzyme loci, respectively). Hence, changes in background levels of antibiotic resistance affect the species as a whole and seem likely to be a response to increased exposure to the compounds over the past several decades. In addition, given that the increase was observed for all of the antibiotics tested, it was presumably caused by refinement of loci conferring nonspecific low-level resistance to multiple antibiotics.

If environmental exposure to antibiotics can elevate levels of the antibiotic resistance, then host populations should directly influence antibiotic resistance profiles. Consistent with this reasoning, the background levels of antibiotic resistance for commercially applied antibiotics are higher in E. coli strains, which are commensals in humans and domesticated mammals, than in Salmonella strains, which circulate principally in nonmammalian populations. In addition, it is noteworthy that Salmonella strains are, on average, more resistant to protamine, a small antimicrobial peptide whose analogs have been recovered from a wide range of vertebrate hosts. However, within E. coli, there is not a significant difference in the antibiotic resistance profiles of strains isolated from human and nonhuman hosts; however, strains of this species are not typically restricted to particular host species or geographic locations (23, 24, 26).

Bacteria have a wide array of mechanisms to protect the cell from destruction by an antimicrobial agent: (i) the antibiotic molecule can be modified or sequestered by genetically encoded enzymes or cellular proteins (2, 15, 27); (ii) bacteria employ active efflux mechanisms and/or permeability barriers to certain compounds (20); and (iii) the specific target of the antibiotic may be modified to make it insensitive to the drug (28). The sporadic occurrence of antibiotic resistance determinants among enteric bacteria suggests that the very high levels of resistance to certain antibiotics are due to a combination of these strategies and are transmitted by mobile genetic elements. For example, high-level resistance to tetracycline and high-level resistance to -lactams are each known to be conferred by at least two mechanisms, which could account for their prevalence in the ECOR and SARC collections, respectively.

Despite the broad distribution of high-level antibiotic resistance in contemporary populations of E. coli and Salmonella, early characterizations of the plasmid contents of pre-antibiotic era strains provided no evidence of transferable antibiotic resistance determinants (13, 16). Although salmonellae do not habitually reside in hosts treated with antibiotics and are likely to have experienced different selective pressures for resistance than E. coli strains have experienced, R plasmids and other genetic elements conferring antibiotic resistance can be efficiently maintained and disseminated within this species by conjugation, transformation, and transduction (4, 19, 21). Our future work will be directed towards identifying the specific genes or alleles that are responsible for the differences in the antibiotic resistance profiles observed within and among these organisms and to examine the MICs of additional antibiotics in other natural populations of enteric bacteria.

Application of antibiotics over the past 50 years has resulted in an unremitting increase in the numbers of commensal and pathogenic bacteria that are resistant to antimicrobial compounds. We have shown that aside from the high frequency of resistance to a particular antibiotic, contemporary populations of enteric bacteria display elevated tolerance in their nonspecific responses to several antibiotics. Although the increases in the background levels of resistance do not threaten control of these organisms, the results show that bacteria, even those not regularly or directly subjected to antibiotic challenge, have changed in response to increases in the application of antibiotics over the past several decades.