Antimicrobial resistance and virulence: a successful or deleterious association in the bacterial world?

Abstract

Hosts and bacteria have coevolved over millions of years, during which pathogenic bacteria have modified their virulence mechanisms to adapt to host defense systems. Although the spread of pathogens has been hindered by the discovery and widespread use of antimicrobial agents, antimicrobial resistance has increased globally. The emergence of resistant bacteria has accelerated in recent years, mainly as a result of increased selective pressure. However, although antimicrobial resistance and bacterial virulence have developed on different timescales, they share some common characteristics. This review considers how bacterial virulence and fitness are affected by antibiotic resistance and also how the relationship between virulence and resistance is affected by different genetic mechanisms (e.g., coselection and compensatory mutations) and by the most prevalent global responses. The interplay between these factors and the associated biological costs depend on four main factors: the bacterial species involved, virulence and resistance mechanisms, the ecological niche, and the host. The development of new strategies involving new antimicrobials or nonantimicrobial compounds and of novel diagnostic methods that focus on high-risk clones and rapid tests to detect virulence markers may help to resolve the increasing problem of the association between virulence and resistance, which is becoming more beneficial for pathogenic bacteria.

Fig 1

Data from in vitro competition experiments with E. coli MG1655 carrying the recombinant plasmids pBGS18-TEM1, pBGS18-SFO1, pBGS18-AmpR-SFO1, pBGS18-OXA10-like, pBGS18-FOX4, and pBGS18-CTX-M-32. The strains expressing AmpR-SFO1, OXA10-like, and OXA24 β-lactamases showed significant fitness costs. The CI values obtained with different β-lactamases are plotted and are representative of eight different experiments, with the median CI values shown by horizontal lines. (Adapted from reference .)

Fig 2

Partial structures and descriptions of different plasmids carrying virulence factors and antimicrobial resistance determinants together. Light gray, virulence genes; dark gray, antimicrobial resistance genes; black, plasmid maintenance genes. (A) Plasmid pMMA2, isolated from clinical isolate AbH12O-A2, which caused a large nosocomial outbreak. (Adapted from reference .) (B) Plasmid pEK499, which carries up to 10 resistance genes, including the genes encoding resistance to β-lactamases (bla_TEM-1, bla_CTX-M-15, and bla_OXA-1), aminoglycosides (aac6-Ib-cr), macrolides [mph(A)], chloramphenicol (catB4), tetracycline [tet(A)], streptomycin (aadA5), and sulfonamide (sulI). It possess two copies of the vagD-vagC virulence-associated system and the toxin/antitoxin pemI-pemK and ccdA-ccdB systems, which are involved in plasmid maintenance by postsegregation killing processes. (Reprinted from reference .) (C) Plasmid pRSB107, which carries genes encoding resistance to the following antibiotics: β-lactams (bla_TEM-1), aminoglycosides (aph), streptomycin (strA-strB), sulfonamide (sulI), macrolides [mph(A)], trimethoprim (dhfR), chloramphenicol (catB4), and tetracycline [tet(A)]. This plasmid carries four putative virulence-associated determinants: an aerobactin iron acquisition siderophore system (iucA, iucB, iucC, iucD, and iutA), a putative high-affinity Fe²⁺ uptake system (orf82), an sn-glycerol-3-phosphate transport system (ugpB, ugp, ugpE, and ugpA), and two copies of the virulence-associated genes vagC-vagD. Approx., approximately. (Reprinted from reference .)

Fig 3

Functional role of bacterial multidrug efflux pumps in natural microbial ecosystems. (Reprinted from reference with permission from John Wiley and Sons.)

Fig 4

Results of in vivo competition experiments in a mouse model of systemic infection (median CI values are shown by horizontal lines). The figure reflects the competition index values of the acrA and tolC efflux pump component mutant strains compared with the wild-type EcD64 and JC194 E. cloacae clinical isolates. Decreased fitness is observed when the efflux system is inactivated. (Adapted from reference .)

Fig 5

(A) Relationship between resistance, fitness, and compensatory mutations. The figure reflects the fitness (measured as growth capacity) of the S. pneumoniae wild-type (WT) strain (1974), the linezolid-resistant mutant (1974M1), the T-7 mutant (the 1974 WT strain with the main mutation involved in resistance, the 23S rRNA mutation G2576), which is less fit than strain 1974M1 (indicating the existence of compensatory mutations in the latter), the T-7^spr1021 mutant (T-7 mutant with overexpression of the spr1021 gene, encoding an ABC transporter), which displays increased resistance to linezolid but with a clear biological cost, and finally the T-7^{spr0333,spr1021,spr1887} mutant (T-7 mutant with overexpression of the spr1021 and spr1887 genes and also the methyltransferase gene spr0333), which appears to restore the original fitness by the action of the new ABC transporter encoded by spr1887 (those commented on in the text are highlighted). (Reprinted from reference , which was published under an open-access license.) (B) High-level resistance to streptomycin in mutants of S. enterica subsp. enterica serovar Typhimurium. The fitness costs are reflected on the y axis. The K42R mutation in the protein encoded by rpsL gene is free of any fitness cost; however, the appearance of the K42N mutation in the same protein is associated with a biological cost, which can be almost totally compensated for by the secondary mutation K205T in RpsD protein. (Modified from reference by permission from Macmillan Publishers Ltd.)

Fig 6

Theoretical model proposed by Handel et al. of the emergence and persistence of antimicrobial resistance in different ecological niches. The bars indicate the fitness of the strains in a given situation, the solid arrows indicate acquisition of mutations that occur frequently due to the large size of the original population, and the dashed arrows indicate acquisition of mutations that occur infrequently in a small part of the original population. If the resistant mutant is less fit than the wild-type strain, it will tend to extinction by competition; however, if during competition the resistant strain develops compensatory mutations, it could eventually emerge and persist. The antibiotic levels in the environment will largely determine this selection. (A) In an environment without antibiotic treatment (e.g., seas), resistant strains are associated with low fitness and will not emerge. (B) In an environment with a certain level of antibiotics (e.g., during infection in a compartment of the treated patient with a low level of antibiotics or without antibiotic distribution), the fitness of the wild-type strain is reduced, which enables emergence of a resistant mutant. One frequent conversion and two infrequent conversions are required. (C) In an environment with antibiotics (e.g., a nosocomial environment), resistant mutants readily emerge, and only one infrequent conversion is necessary. (D) In an environment with a massive presence of antibiotics (e.g., during antibiotic treatment), the low fitness of the original population readily enables the emergence and persistence of resistant populations. (Reprinted from reference , which was published under an open-access license.)

Fig 7

(A) Time until emergence of resistance, based on a theoretical model obtained from simulations of the deterministic model of years until emergence of the resistant strains (see reference for more details). Fitness levels of the resistant strains 1, 2, and 3 are 75%, 85%, and 95%, respectively, of those of the susceptible strain in the absence of antibiotics. Vertical lines represent the level of treatment necessary for the fitness of the resistant strains to equal that of the susceptible original strain. The conversion rate of the strains due to the compensatory mutations equals 10⁻¹ (for other rates, see reference 5). (B) Probability that resistance will emerge within 1 year, based on a theoretical model obtained from stochastic simulations. The parameters used are the same as for panel A. (Reprinted from reference , which was published under an open-access license.)

Fig 8

Model describing the signals controlling expression of PhoP-PhoQ-regulated determinants and the interaction between the PhoP-PhoQ and PmrA-PmrB two-component systems, as well as some of the genes and phenotypes governed by the PhoP-PhoQ system. (Reprinted from reference with permission.)

Fig 9

Growth curves and time-kill curves for E. faecalis OG1RF, relA mutant [relA encodes an enzyme responsible for controlling (p)ppGpp metabolism in Gram-positive bacteria], relQ mutant (relQ encodes an enzyme related to constitutive expression of ppGpp in nonstressed cells), and relAQ double mutant strains in the presence of subinhibitory concentrations of vancomycin (386, 390). (A) Growth curves reveal that the relA mutant grows faster than E. faecalis OG1RF (a strain harboring the ebp gene, which encodes endocarditis and biofilm formation-associated pilus operon). However, relQ and relAQ mutants strains display slow or impaired growth in the presence of vancomycin. (B) Time-kill curves reveal that the relA mutant survives better than the relQ and relAQ mutants, which were killed more rapidly by vancomycin. (Reprinted from reference with permission.)

Fig 10

The mexR and nalC mutations decrease C-30 inhibition of P. aeruginosa QS phenotypes. (a) Swarming motility. The rhlR mutant (rhlR encodes the two transcriptional regulators of the acyl homoserine lactone system of P. aeruginosa) was used as a negative control. MeOH was used as a negative control for C-30. (b) Production of pyocyanin (phenazine toxic metabolites). An asterisk indicates statistical significance. A PA14 phz mutant (a strain without phenazine production) was used as a negative control. (Reprinted from reference by permission from Macmillan Publishers Ltd.)