Fit and resistant is a worst case scenario with bacterial pathogens

Pseudomonas aeruginosa is a prevalent nosocomial pathogen (1) and a threat in neonatal and intensive care units, causing deadly infections in ventilator-associated pneumonia or immunosuppressed patients. P. aeruginosa strains causing these infections are frequently multiresistant, and treatment is a nightmare for clinicians (2). P. aeruginosa colonizing the lungs of cystic fibrosis patients cannot be eradicated even after intensive antibiotic therapy. Combinatorial therapies make use of β-lactams, colistin, fluoroquinolone, or aminoglycosides, but resistance to these drugs keeps emerging at high rate, except for colistin. The acquisition of resistance happens via mutations or acquisition of new functions by horizontal gene transfer. The downside for bacteria acquiring resistance is a pay back in fitness cost. Antibiotics target vital cell functions, such as protein synthesis, DNA supercoiling, or integrity of the cell envelope, and compensatory mutations may result in these features not being reliable (3). This is good news and suggests that once the selective pressure of antibiotic treatment ceased, the resistant strains could be outcompeted by strains that are not resistant but fitter, which will be a stumbling block to the spread of multiresistant organisms. However, the work by Skurnik et al. in PNAS (4) demonstrates that a gain of in vivo fitness by P. aeruginosa is due to the lack of the outer membrane protein OprD, and this event correlates acquisition of carbapenem resistance.

What are the usual mechanisms that preserve bacteria from the killing/stasis impact of antibiotics? Resistance can result from the enzymatic modification/degradation of the drug itself such as with P. aeruginosa strains that encode β-lactamases. Resistance can result from modification of the drug target, such as a mutation in the housekeeping RNA polymerase, which is no longer inhibited by rifampicin. However, an important mechanism, which relates to the study by Skurnik et al. (4), is associated with the permeability of the outer membrane of Gram-negative bacteria. This hydrophobic barrier restricts the diffusion of toxic compounds into the cell. Nevertheless, bacteria have channels, called outer membrane proteins (OMPs), through which small molecules diffuse passively, thus preserving exchange with the milieu. Some channels are nonspecific, and the selectivity is based on the size of the molecule and not on its nature. OprF is a major P. aeruginosa porin, which allows diffusion of small ions or small polar nutriments but does not allow a high rate of diffusion because a large number of the OprF porins are in a close conformation. As a result, the overall permeability of the P. aeruginosa outer membrane is restricted. Other porins display specificity. OprP is a phosphate-specific porin, OprB is a glucose/carbohydrate-specific porin, and OprC is likely involved in copper transport (5). Several P. aeruginosa OMPs, e.g., OprM, OprN, and OprJ, have a role in the efflux of toxic compounds. Each of these OMPs assembles into a tripartite supramolecular complex (e.g., MexAB-OprM) forming an efflux pump, which rejects incoming toxic molecules into the extracellular environment (1). A low outer membrane permeability combined with a high capacity to efflux toxic compounds provides an intrinsic mechanism for drug resistance.

An alternative resistance strategy is not to let the drug get inside the cell. The OMP OprD is involved in the transport of basic amino acids, but also of antibiotics of the carbapenem family (6). Several studies reported that clinical P. aeruginosa strains isolated from patients under carbapenem treatment display an oprD mutation. In their study, Skurnik et al. stumbled on OprD but in an unexpected manner (4). By using a P. aeruginosa transposon (Tn) mutant library, and by testing the ability of this collection to colonize the cecum and disseminate to the spleen of infected mice, Skurnik et al. (4) observed a systematic enrichment in the pool of colonizers for strains carrying a Tn insertion in the oprD gene (>40% of the cecum population). Impressively, more than 90% of the population, which disseminated to the spleen, had an oprD mutation (Fig. 1). Such results raise numerous questions. Why, in this context, is there such a high selective pressure toward an oprD mutation and why does the mutation lead to a dramatically increased fitness? Interestingly, this phenomenon seems to be specific for the in vivo situation since no obvious advantage could be observed for an oprD mutant versus a wild-type strain when grown in the laboratory.

Fig. 1.

Selection of P. aeruginosa oprD mutants during GI tract colonization. (A) Colonization of the GI tract by PA14 Tn mutants (green) positively selects oprD mutants (red) that survive in the cecum and disseminate to the spleen, whereas pil mutants are enriched in the cecum (yellow). (B) The mutation in oprD prevents diffusion of carbapenem, amino acids, or carboxylates into the bacterial cell and changes expression of about 150 genes. (C) The phenotypic traits associated with the oprD mutation are shown, and the result is an increase in mortality.

In a previous report, Skurnik et al. (7) showed that in the course of an infection mutants affected in the pil genes have a colonization advantage (Fig. 1) involved in the assembly of type IV pili. However, in PNAS (4), they show that the oprD mutants are perfectly motile, and the gain in fitness has to be related to other oprD-dependent traits. One way to identify such traits is to compare the transcriptomic profile of an oprD mutant with the one of its parent. Skurnik et al. thus use an RNA-seq approach (4), but a quick examination of the data eliminated the usual suspects such as protein secretion systems or biofilm-related functions, which may have accounted for the gain in fitness. This is remarkable because one of the phenotypic traits of the oprD mutant is high cytotoxicity (Fig. 1), which thus appears to be independent of the type III secretion system (T3SS), which is one of the most potent P. aeruginosa weapons for toxicity (8). It should be noted that the transcriptomic analysis was conducted on bacterial cultures grown in vitro and might not identify all of the genes reflecting the fitness advantages observed in vivo.

Skurnik et al. (4) describe other OprD-associated phenotypes that may explain in vivo fitness. In particular, oprD mutants survive better serum-mediated killing and thus host immune response as well as acidic conditions, which could be found in the gastrointestinal (GI) tract (Fig. 1). It is too early to understand the mechanism associated with this phenomenon, and systematically addressing the relevance of any 1 of the 150 genes whose expression varies in the oprD mutant (Fig. 1) might be fastidious. The complexity in identifying the causal effect on fitness is further challenged by the fact that OprD is not only transporting carbapenem but is a channel for transporting small nutriments. It is proposed to belong to the outer membrane carboxylate channel (Occ) family, which is involved in the transport of molecules containing a carboxyl group (9). The lack of transport of amino acids or carboxylates may greatly impact metabolic

The work performed by Skurnik et al. challenges the common concept that resistant bacteria are less fit.

fluxes and numerous intracellular processes, which in turn impact the overall behavior of the bacterium.

The beauty of the study by Skurnik et al. (4) is also reflected by the power of high-throughput sequencing. Whereas several studies have made the observation that oprD mutants were systematically selected as carbapenem-resistant isolates, demonstration of fitness gain, and in this case enhanced GI tract colonization, is a more challenging question to address. Previous strategies to identify important genetic traits during in vivo infections made use of a collection of mutants tagged with specific labels [signature tagged mutagenesis (STM)]. Such studies in P. aeruginosa were conducted with a library of >6,000 STM mutants. Input pools of about 70 mutants for each round were used to identify traits during chronic infections of mouse airways, which prevent or positively influence colonization and persistence (10, 11). In the case of the study of Skurnik et al. (4), it is a random and highly saturated library of >300,000 Tn insertions that is used as an input pool (Fig. 1). The output pool is characterized by sequencing the genomic DNA from the recovered population (Tn-Seq or INSeq), which allows quantification of the transposon inserting at low or high frequency at a certain position in the genome (12) and thus points out deleterious or advantageous genes.

It is remarkable to observe that oprD is the main selected mutation for the gain in fitness in GI tract colonization, and we are only at the preliminary stages in understanding the mechanism that provides such a strong advantage. In the case of P. aeruginosa strains infecting the lungs of cystic fibrosis patients, it is rather clear why certain types of mutations are selected. For example a mucA mutation results in an increase of slimy alginate production or a lasR mutation reduces the production of virulence factors, which overall benefits the chronic mode of infection (13, 14).

The work performed by Skurnik et al. (4) challenges the common concept that resistant bacteria are less fit, and their results may dampen our enthusiasm in finding solutions on how to outcompete pathogens that acquire antibiotic resistance. Until now, an improved fitness has only been shown for carbapenem resistance and could be a feature of the oprD mutation that may not apply to other mechanisms of resistance. Once a resistance emerges, the best strategy is to stop the use of the incriminated drug, but the antibiotic pipeline is quickly drying up. We need to trust in our ability to find new drugs and to address and solve pertinent and difficult questions relevant to bacterial fitness and resistance. The study of Skurnik et al. (4) is a nice example of why we need to move our playground from the Petri dish to the in vivo situation.