Stress-adaptation, co-selection, cross-resistance, and cross-protection
Mechanisms exist whereby microorganisms that are resistant to one antimicrobial may become resistant to others (Yousef and Juneja 2003). Exposure to subinhibitory concentrations of an antimicrobial, for example, may activate intrinsic resistance mechanisms, thereby decreasing susceptibility of the microbe to the inducing agent and in tandem decreasing susceptibility to other, unrelated antimicrobials. In other instances, resistance to several antimicrobials having unrelated targets or modes of action may result from co-selection, which involves sequential linking of separate genes conferring resistance to different antibiotics, often on plasmids or integrons,12 and transferred together. Cross-resistance is the occurrence of resistance to antimicrobials because they have the same molecular targets. Cross-protection occurs when adaptation to one stress is associated with increased resistance to another, unrelated stress. Correlations among these mechanisms are seen in some cases, but the root causes of the dissemination of the resistance remain unknown.
Strains of E. coli resistant to thymol and eugenol (essential oils found in thyme and cloves, respectively) were found to be more resistant to chloramphenicol (Walsh and others 2003). Because stable resistance to the essential oil components was not readily detected, the authors denoted the increased resistance as “tolerance” (Walsh and others 2003). In contrast, methicillin-resistant Staphylococcus aureus, however, were found to be as sensitive to oregano essential oil and its components, carvacrol and eugenol, as methicillin-sensitive strains (Nostro and others 2004). Resistance to carvacrol, however, which is associated with changes in the cellular membrane, apparently does not confer resistance to other membrane-active compounds. Bacillus cereus adapted to carvacrol were demonstrated to be more sensitive to subsequent nisin exposure than nonadapted cells (Pol and others 2001).
Bacteria are able to produce stress response proteins when subjected to subinhibitory levels of stress (Yousef and Juneja 2003). A variety of situations can induce transcription and translation of stress response proteins, which convey increased resistance to a multitude of stressors. For example, exposure of E. faecalis to subinhibitory levels of sodium chloride, sodium dodecyl sulfate, and bile salts conferred a protective effect against heat compared to nontreated cells (Flahaut and others 1997). Heat shock proteins (HSP) comprise one of the most well-studied classes of stress response proteins, although the HSP levels do not correlate with the extent or persistence of protection (Jorgensen and others 1996; Mackey and Derrick 1990). HSP are typically regulated by sigma factors such as RpoS or RpoH, which are subunits of RNA polymerase.
Salmonella enterica serovar Enteritidis and L. monocytogenes first exposed to alkali are more resistant to heat treatment than those not pre-exposed (Humphrey and others 1991; Taormina and Beuchat 2001). Studies with Salmonella Enteritidis showed that treatment with low levels of alkali (pH 10.0 sodium hydroxide or trisodium phosphate) resulted in a decrease in protein expression of 15% and 22%, respectively (Sampathkumar and others 2004). Some outer membrane proteins, identified as protein chaperones and housekeeping proteins involved in biosynthesis, were up-regulated. Similarly, when E. coli K-12 was shifted from pH 7 to 8.8, known HSPs were induced (Taglicht and others 1987).
Hong and others (2002) found that Streptomyces coelicolor, containing a plasmid encoding a signal transduction system including the sigma factor E (ΦE), demonstrated lysozyme-induced resistance to kanamycin (100g/mL). L. monocytogenes has been shown to contain a similar signal transduction system (CesRK) that is activated upon introduction of lysozyme to the cells and results in antibiotic resistance.
An example of an intrinsic resistance system is the multiple antimicrobial resistance (mar) operon, a global regulator that controls intrinsic resistance to unrelated antibiotics and other cytotoxic substances (Alekshun and Levy 1999). Golding and Matthews (2004) demonstrated decreased susceptibility of E. coli O157:H7 to multiple antimicrobials, putatively linked to a mutation in the mar operon, following exposure to chloramphenicol. Potenski and others (2003) found that upon treating Salmonella Enteritidis cells with sublethal levels of chlorine, sodium nitrite, sodium benzoate, or acetic acid, the cells exhibited resistance to tetracycline, chloramphenicol, nalidixic acid, and ciprofloxacin, thus determining that a mar operon was responsible for the resistance responses.
Antimicrobial resistant phenotypes of E. coli O157:H7 may also be related to acquisition of class 1 integrons (Zhao and others 2001a), which is significant because the integrons may contain several antimicrobial gene cassettes and, therefore, co-select for resistance to other antimicrobials.
Genes encoding for multidrug efflux systems in S. aureus have been located on plasmids (generally 18 to 57 kb in size) also containing genes for resistance against penicillin, gentamicin, trimethoprim and kanamycin (Lyon and others 1984). The qacA and qacB genes have been found on plasmids that also confer resistance to various antibiotics, including penicillin (Lyon and Skurray 1987). Twenty-four QAC-resistant Staphylococcus isolates were analyzed for resistance to selected antibiotics and dyes (Heir and others 1999). Five of the seven strains with the QAC resistance genes qacA/qacB had high-level resistance to penicillin G and ampicillin. One isolate containing the smr gene showed resistance to ampicillin, penicillin G, tetracycline, erythromycin, and trimethoprim, but not to chloramphenicol, gentamicin, norfloxacin, kanamycin, or vancomycin. It was suggested that the antibiotic resistance in this strain was due to resistance markers on the chromosome or other plasmids harbored by the strain. All other sanitizer-resistant isolates were generally susceptible to antibiotics.
Several studies have found a lack of cross-resistance between agents, even when mechanisms appear similar. For example, when acquired resistance mechanisms for biocides, which can closely resemble those for antibiotics, were studied by Aase and others (2000), no connection was found between QAC resistance and antibiotic resistance in L. monocytogenes. They evaluated 200 L. monocytogenes isolates from various food, human, and environmental sources from Norway and Europe and found that 10% were resistant to benzalkonium chloride (BC), while none of the isolates was resistant to any of the 15 antibiotics. Both resistant and sensitive strains responded approximately equally to BC after adaptation, and remained stable during subculturing in the absence of BC. They suggested that genes coding for the efflux pumps providing resistance against QAC and ethidium bromide are not located on the multiple drug resistance (MDR) plasmid. When sublethal levels of a triclosan-containing domestic detergent were applied to a biofilm, the composition of the biofilm changed; however, the remaining organisms were generally as susceptible to a host of antibiotics and other antimicrobials as the initial population (McBain and others 2003).
There are many other instances where resistance to one antimicrobial does not confer resistance to another. This can often be explained by a mechanistic understanding of the agent's effect on the cell. Enterococci are particularly resistant to heat and sodium hypochlorite (Freeman and others 1994; Kearns and others 1995), which may permit their survival of intervention techniques in both food processing and clinical settings. In one study, vancomycin-resistant enterococci did not have enhanced resistance to chemical disinfectants or to heat (Bradley and Fraise 1996). This was confirmed by Panagea and Chadwick (1996), who found no differences in heat tolerance of vancomycin-resistant or sensitive clinical isolates of E. faecium.
Bertolatti and others (2001) and Walsh and others (2001a) reported that the potential for antibiotic-resistant organisms to exhibit enhanced resistance to food preservation techniques or food antimicrobial agents has been studied only to a limited extent. Antibiotic-resistant Gram-positive cocci and streptomycin-resistant L. monocytogenes responded similarly to heat compared with the corresponding wild-type strains. Other investigations have considered the decimal reduction times (D-values) of antibiotic-resistant organisms with or without induced acid tolerance to determine whether heat resistance is altered in the strains. For example, Bacon and others (2003a) examined thermal D-values of wild type and MDR-Salmonella strains isolated from bovine sources and grown in various levels of glucose to stimulate an acid tolerance response (ATR). At 59 oC, acid-tolerant cultures had increased thermal resistance compared with nonacid tolerant controls. The D61° C values of antimicrobial susceptible Salmonella strains increased as the glucose concentration (acid tolerance) in the culturing medium increased, but D61o C values of MDR-cultures were similar, irrespective of ATR. When averaged across glucose levels and temperatures, D-values of antimicrobial susceptible and resistant Salmonella cultures were similar. The results suggest a cross-protective effect of acid adaptation on thermal inactivation, but no association between antimicrobial susceptibility and heat resistance. When the ATR was induced in either type of strain by growth in glucose, some strain variations in acid resistance were observed, but no association between susceptibility to antimicrobial agents and potential to survive a low pH stress was made (Bacon and others 2003b). Lopes (1998) reported observing that antibiotic-resistant strains of Salmonella Typhimurium and L. monocytogenes were equally as susceptible to sanitizer treatments as antibiotic sensitive strains. They found that Salmonella Typhimurium strains resistant to nalidixic acid and L. monocytogenes strains carrying plasmid pGK12 encoding resistance to chloramphenicol, erythromycin, and rifampin did not exhibit resistance to organic acid/anionic surfactant-based sanitizers. Others have concluded that in Listeria spp., including the pathogenic L. monocytogenes, plasmid-mediated disinfectant resistance may not necessarily be linked to antibiotic resistance (Lemaitre and others 1998).
Mazzotta and others (2000) found that nisin-resistant L. monocytogenes and C. botulinum were not more sensitive to food preservatives such as low pH, salt, sodium nitrite, and potassium sorbate. Hossack and others (1983), however, reported that nisin resistance in S. aureus was linked with antibiotic resistance, observing that antibiotic MICs increased as much as 30-fold among the nisin-resistant strains. Several studies suggest that nisin resistance results in physiological changes that decrease resistance to other agents. These data are not necessarily inconsistent, as different antibiotics have different modes of action, which may or may not be affected by the changes in membranes. Szybalski (1953) reported that a penicillin-resistant S. aureus mutant was 50 times more sensitive to nisin. Severina and others (1998), however, found that several MDR-bacteria remained sensitive to nisin treatment. Similarly, studies with nisin-resistant L. monocytogenes or cells pretreated with nisin showed no significant increase in resistance to antibiotics (Crandall and Montville 1998). McEntire (2003) observed that the nisin-resistant strain was highly sensitive to second and third generation cephalosporins, at concentrations where the wild type was virtually unaffected. The mutant also exhibited increased acid sensitivity due to increased ATPase activity; while acid sensitivity may not be directly related to nisin resistance, both phenotypes may be directly or indirectly controlled by the same signal transduction system (Cotter and others 2002; McEntire and others 2004).
“Collateral sensitivity” (a mutation or adaptation conferring resistance to one or more agents which simultaneously increase sensitivity to other agents) is not unique to nisin resistance. Bacillus licheniformis, which is resistant to the bacitracin it produces, is highly sensitive to detergents, likely due to a specific membrane change (Podlesek and others 2000). After exposure to alkali cleaning solutions, 4 of 5 strains of L. monocytogenes were as sensitive or more sensitive to heat than unexposed cells, and all were more sensitive to the sanitizer components (free chlorine, benzalkonium chloride [BC], and cetylpyridinium chloride) compared with the controls (Taormina and Beuchat 2002).
Development of resistance to acid and heat among pathogens may influence their behavior when exposed to fermentation, drying, cooking, or consumption in the human host. The increased virulence may stem from the influence of acid resistance on microbial behavior upon exposure to the final barrier (gastric secretions, phagocytosomal vacuoles) in the human host. Thus, in addition to increased resistance against food preservation treatments, stress-adaptation may lead to increased virulence and lower infectious doses (Samelis and Sofos 2003a).
Stopforth and others (2004a), however, indicated that similarly acid-adapted (glucose) E. coli O157: H7 inocula were not different than controls in survival when inoculated in wounds of apples and exposed to water or sanitizing solutions of acetic acid, hydrogen peroxide, and sodium hypochlorite. Ikeda and others (2003) found no differences in survival or growth of acid-adapted (glucose) L. monocytogenes inocula on fresh beef decontaminated with hot water and organic acid solutions. Calicioglu and others (2002a, 2002b, 2003a, 2003b, 2003c, 2003d) reported that inactivation of acid-adapted (glucose) inocula during drying and storage of beef jerky was more efficient than that of normal cultures (grown in broth without glucose), potentially indicating that exhaustion or stressing of the cells during acid adaptation caused the cultures to be more sensitive to the subsequent stresses of acid, heat, and dehydration, and confirming the importance of the hurdle concept in food preservation.
As is the case for other pathogens, L. monocytogenes, which can grow at a pH as low as 4.39 (George and others 1988), can exhibit the ATR with increased survival of prestressed cells at normally lethal acid levels (Bonnet and Montville 2005; Gahan and others 1996; Samelis and others 2003). This adaptive mechanism, which may occur in different pH ranges for different microorganisms (Koutsoumanis and Sofos 2004b), does not enhance the ability of the organism to grow, but has several implications for food safety due to the increased pathogen survival rates. For example, Bonnet and Montville (2005) showed that ATR-induced L. monocytogenes coinoculated in broth with a nisin-producing Lactococcus lactis persisted in the majority of samples for at least 30 days. L. monocytogenes that were not induced to ATR, however, could not be detected. Cross-protection of acid tolerance in L. monocytogenes with thermal tolerance, crystal violet, ethanol, and osmotic stress has also been demonstrated (O'Driscoll and others 1996).
Since L. monocytogenes must be able to bypass the acidity of the stomach in order to be infective, the impact of ATR induction on microbial survival with preexposure to acids or other stressing antimicrobial hurdles in simulated gastric fluid has been of interest. Results indicate that simulated gastric fluid acid tolerance may depend on type and composition of product, microbial cell concentration, cell age, and product storage time (Stopforth and others 2005). The authors noted observing, for example, higher gastric fluid ATR with increased product fat content. However, spontaneous mutants of L. monocytogenes with constitutive acid tolerance showed increased virulence in mice when administered intraperitoneally, suggesting that a mechanism in addition to gastric acid resistance is involved (O'Driscoll and others 1996).
Dissemination of resistance determinants between microorganisms
Two main factors contribute to the persistence of antimicrobial resistant microorganisms in the environment: survival of the microorganism and maintenance of the resistant genotype. Dissemination of resistance determinants can occur at three levels—bacterial (clonal spread), replicon (plasmid epidemics), or gene (transposons), all three of which coexist in nature and are not only infectious but exponential as well, since all are associated with DNA duplication (Courvalin 2005).
The “mobility” of these antibiotic resistance genes is attributed to their residence on mobile genetic elements—plasmids (Navarro and others 2001; Smalla and others 2000), transposons (Sundin 2002), and integrons (Nandi and others 2004), described in detail in Appendix 2. Gene transfer between pathogens is not a new concern and has been reported in pathogens of both humans and animals. Although the existence of mobile genetic elements predates the widespread use of antibiotics (Hughes and Datta 1983), current problems have arisen because more and more resistance genes have become linked in multiple, tandem repeats in these mobile DNA elements.
Starliper and others (1998) examined strains of E. ictaluri resistant to sulfadimethoxine/ormetoprim and found that resistance to sulfadimethoxine/ormetoprim and tetracycline was carried on a 55 kb R-plasmid. The R-plasmid allowed very fast and efficient transfer of resistance between E. ictaluri and E. coli and vice versa. Although the origin of the plasmid was unknown, it was found to be essentially identical to a plasmid found in Tribrissen®-resistant E. coli strain 1898 originating from a case of equine cystitis (Cooper and others 1993). The implication was that antibiotic resistance found in the fish pathogen could possibly have originated with bacteria colonizing warm-blooded animals.
Chee-Sanford and others (2001), possibly the first group to use DNA technology to study the genes for a major class of antibiotic resistance in groundwater potentially impacted by animal agriculture, used PCR typing methods to assess the presence of tetracycline resistance determinants in waste lagoons and groundwater underlying two swine farms impacted by waste seepage. All eight classes of genes (tet(O), tet(Q), tet(W), tet(M), tetß(P), tet(S), tet(T), and otr(A) encoding this mechanism of resistance were found in total DNA extracted from water from both lagoons. The authors noted that the maximal relative frequency and diversity of tetracycline resistance genes occurred at waste lagoons and gradually declined in the direction of groundwater flow; however, one of the genes was still detectable 250 meters downstream.
Agerso and others (2004) studied the presence of the tet(M) gene in farmland soil by direct detection of the gene. They reported that the gene was most prevalent in farmland soil immediately after spread of pig manure slurry, but could be detected on farmland soil 2 y after the field had been treated. On soil not treated with animal manure, tet(M) could only be detected after selective enrichment with tetracycline present in the media under anaerobic and aerobic conditions. The results indicate that the tet(M) gene is spread with bacteria in the manure, but that it is also present in the indigenous soil microflora, possibly occurring specifically in the facultative anaerobic bacteria. Sengelov and others (2003) investigated the level of tetracycline, erythromycin, and streptomycin resistance among bacteria before and after spread of pig manure slurry on fields. They found that the ratio of colony forming units (CFU) of tetracycline-resistant bacteria to all bacteria was significantly higher immediately after spread of pig manure slurry. The ratio decreased rapidly 1 y after the spread, showing no accumulation of tetracycline-resistant bacteria. No effect on erythromycin- and streptomycin-resistant bacteria in farmland soil was observed in the study.
Another means of environmental transfer of antibiotic resistance genes from the antibiotic-producing strain might be through direct ingestion of medicated feeds by food animals. It has been shown that a DNA-encoding homolog of the van resistance gene cluster was a contaminant of feed-grade avoparcin. Thus, it was proposed that the ingested glycopeptide resistance gene complex was present, conferred resistance to this antibiotic, and in the presence of the selective pressure of the glycopeptide avoparcin in the food animal, selected for increased numbers of resistant strains (Lu and others 2004; Marshall and others 1998). However, another study based on amino acid sequence homology showed that horizontal transfer to human or animal bacteria of antibiotic resistance genes from bacteria that are used for antibiotic production was unlikely (Lau and others 2004).
Transfer to humans from various sources
Data on the transfer of resistant organisms from animal or environmental microbial isolates to humans, ability to cause illness, and resultant treatment failure are valuable for assessing the overall impact of antimicrobial resistance on human health. The transfer of antimicrobial-resistant bacteria from food animals to humans is well documented (Sanchez and others 2002; Swartz 2002). Evidence includes transfer of Salmonella from cattle, chickens, pigs, and turkeys (Angulo and others 2000; Mead and others 1999; Meng and others 1998) and Campylobacter species from chickens and turkeys in a commercial operation (Altekruse and others 2002). Farmers may be at a greater occupational risk of acquiring antimicrobial resistant bacteria from the environment. A range of microorganisms including S. aureus, nongroupable Streptococci, Enterobacter, Enterococci, and E. coli isolated from farm workers was significantly more resistant to most antimicrobials than isolates from nonfarm workers (Aubry-Damon and others 2004).
Most ceftriaxone-resistant Salmonella infections are acquired outside the United States. A domestically acquired ceftriaxone-resistant Salmonella infection, however, was reported in a 12-year-old child (Fey and others 2000; Herikstad and others 1997). The ceftriaxone-resistant Salmonella enterica isolate that infected the child was indistinguishable from one of the ceftriaxone-resistant isolates present in a herd of cattle during an outbreak on the family ranch. Although use of ceftriaxone or other antimicrobials in the herd could not be established, it was suggested that use in the herd of ceftiofur, a broad-spectrum cephalosporin approved for use in cattle, most likely led to the emergence of resistance in the S. enterica in these cattle, transmission of the resistant strain from the cattle to the child, and illness in the child (Fey and others 2000). The means of transmission of the ceftriaxone-resistant Salmonella from the cattle to the child was not known; however, Fey and others (2000) thought it unlikely that the child's infection was actually foodborne. They concluded that the inoculum of Salmonella necessary to cause illness in the child might have been lowered by the prior treatment of the child with amoxicillin-clavulanate and ampicillin-sulbactam.
Resistant bacteria on food crops destined for consumption by humans may provide a route of delivery of resistance genes to the human intestinal flora. Enterobacteriaceae are not only found in abundance in the environment, but as pathogens as well as commensals in the human gastrointestinal tract. For example, the same serotypes of E. coli and Klebsiella were found in food served in a hospital setting and isolates of consuming patients (Cooke and others 1970; Cooke and others 1980). A Finnish study investigated the potential for raw vegetables to serve as a source of resistant strains of Enterobacteriaceae (Osterblad and others 1999); researchers concluded that bacteria from vegetables were not responsible for the high prevalence of resistant Enterobacteriaceae in fecal flora. More research is warranted to determine the impact of antimicrobial resistant environmental commensal bacteria as an important source of resistance in fecal flora.
Cross-species infections between plants and humans are increasingly recognized (Tan 2002; Vidaver 2005). Pseudomonas aeruginosa, Burkholderia (Pseudomonas) cepacia, and Serratia marcescens, which can be plant pathogens, are potential serious human pathogens. The plant pathogens are intrinsically antibiotic resistant (Vidaver 2005). However, as yet, there are no data that indicate transfer of antibiotic resistance determinants from the plant pathogens to bacteria causing human disease, or vice versa, under natural conditions (McManus and others 2002).
Three hundred species of fungi have been reported as causing cutaneous and invasive human infections (Taylor and others 2001). The level of invasive infections is attributed in part to increased organ transplants and attendant immunosuppression, as well as complications arising from AIDS, although fungal diseases are reported in “normal individuals” as well (Ponton and others 2000). The human health concern is that some of the bacteria and many of the fungal taxa long known as plant pathogens are being isolated from human infections (Vidaver 2005).
The significance of antibiotic use in domestic aquaculture to food safety and human health is unknown. Ultimately, data relating to the persistence of antibiotic residues and bioactivity in the fish farm environment and the ability of fish pathogens to transmit antibiotic resistance determinants to human pathogens will be required. Most fish pathogens do not infect humans because they are incapable of growing at human body temperatures; thus, the risk of transmission of pathogens from fish to humans is very small. So far, the potential seems more likely for human or animal pathogens to transmit resistance to fish pathogens. Currently, antibiotic usage in aquaculture is at its lowest point since the early 1980s, and until new drugs are approved, the situation seems unlikely to change.