Species-specific mutation rates for ampC derepression in Enterobacterales with chromosomally encoded inducible AmpC β-lactamase

Species-specific mutation rates for ampC derepression in Enterobacterales with chromosomally... Abstract Background AmpC β-lactamases are encoded on the chromosomes of certain Enterobacterales and lead to clinical resistance to various β-lactams in case of high-level expression. In WT bacteria with inducible AmpC, the expression is low, but selection of stably ampC-derepressed mutants may occur during β-lactam therapy. Thus, for Enterobacter spp., Citrobacter freundii complex, Serratia spp. and Morganella morganii that test susceptible in vitro to oxyimino-cephalosporins, the EUCAST expert rules recommend suppressing susceptibility testing results for these agents or noting that their use in monotherapy should be discouraged, owing to the risk of selecting resistance. However, clinical observations suggest that emergence of resistance is not equally common in all species with inducible AmpC. Objectives To determine species-specific mutation rates, which are more accurate and reproducible than previously described mutant frequencies, for ampC derepression in Enterobacterales with inducible AmpC. Methods Mutation rates were determined using a protocol based on Luria–Delbrück fluctuation analyses. Overall, 237 isolates were analysed. Results Mutation rates were high in Enterobacter cloacae complex, Enterobacter aerogenes, C. freundii complex and Hafnia alvei isolates, with a mean mutation rate of 3 × 10−8. In contrast, mean mutation rates were considerably lower in Providencia spp., Serratia spp. and especially M. morganii isolates. Furthermore, we observed species-specific variations in the resistance patterns of ampC-derepressed mutants. Conclusions Our data might help to predict the risk of treatment failure with oxyimino-cephalosporins in infections by different Enterobacterales with inducible AmpC. Moreover, we make a proposal for optimization of the current EUCAST expert rule. Introduction Inducible AmpC β-lactamases are clinically important enzymes encoded on the chromosomes of certain Enterobacterales, mainly including Enterobacter cloacae complex, Enterobacter aerogenes, Citrobacter freundii complex, Hafnia alvei, Providencia spp., Serratia spp. and Morganella morganii.1 They hydrolyse penicillins, cephalosporins and monobactams. In WT bacteria, AmpC expression is low but inducible (via a complex regulation mechanism) in response to β-lactam exposure, and mediates clinical resistance to β-lactams that are strong inducers (cephamycins such as cefoxitin, first/second-generation cephalosporins such as cefuroxime, and aminopenicillins±β-lactamase inhibitors) but not to β-lactams that are weak inducers (e.g. oxyimino-cephalosporins).1 However, mutations may occur spontaneously, resulting in stable derepression of the ampC gene.1,2 The subsequent constant high-level expression of AmpC leads to clinical resistance to all β-lactams apart from carbapenems and sometimes cefepime.1 In infections by WT bacteria with inducible AmpC, therapy with, e.g. oxyimino-cephalosporins can select for pre-existing stably ampC-derepressed mutants, resulting in breakthrough infection and treatment failure. Thus, for Enterobacter spp., C. freundii complex, Serratia spp. and M. morganii testing as susceptible in vitro to oxyimino-cephalosporins, the EUCAST expert rules recommend suppressing susceptibility testing results for these agents or noting that their use in monotherapy should be discouraged, owing to the risk of selecting resistance.3 Of note, clinical observations suggest that emergence of resistance is more common in Enterobacter infections (8%–19%) than in infections by other species (e.g. 3% in Citrobacter infections),4–6 but data are scarce. Knowing exact species-specific mutation rates for ampC derepression would help to predict the risk of treatment failure with oxyimino-cephalosporins in infections caused by Enterobacterales with inducible AmpC because they would allow estimation of the extent to which spontaneously arising mutants are present. To the best of our knowledge, species-specific mutation rates for ampC derepression have not been described for Enterobacterales with inducible AmpC. In previous studies, the authors merely determined mutant frequencies by simply counting the number of mutant bacteria that had grown on selective media in relation to the number of plated bacteria.7–12 Others utilized serial passaging on selective media to compare different antibiotics for their resistance selection potential.13,14 However, the results gained with both methods are inadequate indicators of spontaneous mutation rates. Luria–Delbrück fluctuation analyses are the correct method for determining mutation rates, as outlined by Rosche and Foster.15 Thus, in the present study, we performed Luria–Delbrück fluctuation analyses, which have not been applied to AmpC mutational resistance studies before, in order to provide species-specific mutation rates for ampC derepression. As Rosche and Foster15 clearly and extensively explain the theoretical basis and the methodological details, we recommend reading their paper for detailed background information on the method. Briefly, mutant frequencies (i.e. the proportion of cells in a population that are mutants) are inherently inaccurate because the population of mutants is composed of clones, each of which arose from a cell that sustained a mutation. The size of a given mutant clone (i.e. the number of its individual bacteria) will depend on when during the growth of the population the mutation occurred. Consequently, considering that bacteria grow exponentially and mutations arise stochastically, the fraction of mutants is largely influenced by the timepoints at which the mutations occurred and not only by the number of mutations (i.e. the mutation rate). Hence, even if results from replicate cultures are averaged, mutant frequencies will not be a reliable and reproducible measure of the mutation rate. The distribution of mutant numbers among replicate cultures is usually referred to as the Luria–Delbrück distribution. With the help of this theoretical distribution, the mutation rate can be mathematically estimated using fluctuation analyses. A fluctuation test begins with the inoculation of a small number of cells into a large number of parallel cultures. The cultures are allowed to grow to saturation in non-selective media, and then each culture is plated on selective media for detection of the mutants that have occurred. The observed distribution of the numbers of mutants among the parallel cultures is then used to calculate the probable number of mutations per culture that gave rise to that distribution. Finally, the number of mutations is converted to the mutation rate by dividing it by some function of the total number of cells at risk. Materials and methods Specimens Randomly selected non-duplicate E. cloacae complex, E. aerogenes, C. freundii complex, H. alvei, Providencia rettgeri, Providencia stuartii, Serratia marcescens, Serratialiquefaciens and M. morganii isolates were collected between January 2016 and July 2017 during routine diagnostics in our affiliated clinical microbiology laboratory, provided that they tested as susceptible in vitro to oxyimino-cephalosporins (i.e. MIC ≤1 mg/L for cefotaxime and ceftazidime). For improved readability, these isolates are subsequently referred to as WT strains, although we did not consider whether or not resistance to other antibiotics was present. Species identification was achieved by MALDI-TOF (Microflex LT system, Bruker Daltonics; MBT Compass 4.1.31 software), and antimicrobial susceptibility testing was done according to EUCAST recommendations,16 mainly using Vitek®2 (bioMérieux). Fluctuation assay The fluctuation assay protocol was established according to the instructions of Rosche and Foster.15 For each WT isolate, multiple parallel cultures, each comprising 100 μL of a 105 cfu/mL organism suspension in Mueller–Hinton Broth (Oxoid), were incubated overnight at 37°C. The number of parallel cultures is a critical factor: if the mutation rate is very low, no mutants will arise in the parallel cultures that can be detected during the subsequent selection step, unless an extremely large number of cultures is used. Yet the occurrence of at least one mutant in at least one culture is the precondition that a mutation rate can be calculated. As a consequence, the minimal mutation rate that can be determined (i.e. the limit of detection) is dependent on the number of parallel cultures and the bacterial count per culture. We adjusted the number of parallel cultures dependent on the species, as evaluated in preliminary experiments. For Enterobacter spp., C. freundii complex and H. alvei, 16 cultures were sufficient to gain mutants in almost all WT isolates (otherwise, the experiment was repeated with 32 cultures to lower the limit of detection). For Providencia spp. and Serratia spp., we used 32 cultures, which enabled us to find mutants in a considerable number of isolates. For M. morganii, we were not able to detect any mutants using 32 cultures and, thus, extended experiments to 96 cultures. Two cultures were used to determine bacterial counts in duplicate by plating serial dilutions on non-selective media, i.e. each dilution was mixed with 8 mL of Mueller–Hinton Soft Agar (Oxoid) and then dispersed in a culture plate. The remaining cultures were plated on selective media, i.e. each entire culture was mixed with 8 mL of Mueller–Hinton Soft Agar containing cefotaxime (8 mg/L; Oxoid) and then dispersed in a culture plate. Cefotaxime is, as a weak inducer of AmpC, suitable for selection of ampC-derepressed mutants.1,12,17 Culture plates were incubated overnight at 37°C, then the colonies in/on the soft agar were counted for determination of bacterial counts and mutant numbers per parallel culture. As AmpC inducibility by β-lactams varies between genera,17 in a few strains background growth (induced WT) complicated mutant counting. Then, the experiment was repeated with another weak inducer, namely piperacillin (32 mg/L; Oxoid),1,12,17 to confirm results. In the re-tested strains, using piperacillin reduced background growth, which enabled us to gain reliable mutant counts. However, we could not generally replace cefotaxime by piperacillin for mutant selection in this study because of isolated resistance to piperacillin in some WT strains, possibly because of an additional narrow-spectrum β-lactamase such as TEM or SHV. Cefotaxime and piperacillin concentrations were chosen considering the MIC distributions published on the EUCAST web site. Based on the distribution of mutant numbers in the parallel cultures, the probable number of mutations was estimated using the MSS maximum likelihood method. Then, mutation rates per cell per generation were calculated from the number of mutations and the bacterial count, applying the In(2) correction as suggested by Armitage.18 The formulas used for the calculations are described in detail in the publication by Rosche and Foster.15 For isolates that yielded no mutants (subsequently referred to as mutant-negative isolates, although mutants might have arisen in case of a larger number of parallel cultures), a mutation rate could not be calculated. Thus, we calculated the limit of detection instead, by assuming that exactly one mutant was found in one parallel culture. Phenotypic characterization of WT strains and mutants Disc diffusion using Oxoid discs was done for WT strains and mutant colonies according to EUCAST recommendations.16 In addition, for the WT strains, we checked for presence of an inducible AmpC by a disc induction test, placing strong inducers (cefoxitin and imipenem) adjacent to weak inducers (cefotaxime, ceftazidime and piperacillin). To confirm ampC derepression as the underlying resistance mechanism, mutants were subjected to the boronic acid inhibition test as boronic acid is a well-known inhibitor of AmpC.1 The test was performed by the standard disc diffusion method, using cefotaxime discs alone and in combination with 300 μg of 3-aminophenylboronic acid (APB; ACROS Organics). We tested seven representative mutant colonies from different parallel cultures, or all mutants if fewer than seven cultures yielded mutants. Results We established a fluctuation assay protocol for calculation of mutation rates for ampC derepression for Enterobacterales with inducible AmpC. The number of WT isolates analysed per species, the mutation rate for each isolate and mean mutation rates per species are shown in Figure 1. With the selected number of parallel cultures, we observed mutants in 126/132 (95%) Enterobacter spp., C. freundii complex and H. alvei WT isolates (of note, in the six mutant-negative isolates, the AmpC was not inducible in the disc induction test). In the other species, mutants occurred in only 33/80 (41%) Providencia spp. and Serratia spp. isolates, and in only 3/25 (12%) M. morganii isolates despite the higher number of parallel cultures. Figure 1. View largeDownload slide Species-specific mutation rates. Mutation rates were determined for each strain as described. Results are shown grouped according to species. Each symbol represents one strain. Black colour indicates that mutants were detected and grey colour indicates that mutants did not occur. In the latter case, the limit of detection is given, which was calculated for each mutant-negative isolate by assuming that one single mutant was found (i.e. it is dependent on the number of parallel cultures and the bacterial count). In addition, the number (n) of tested isolates per species is given, and the mean mutation rates, which were calculated per species based on all calculated mutation rates and limits of detection are depicted as horizontal bars and also at the top of the figure for each organism. Figure 1. View largeDownload slide Species-specific mutation rates. Mutation rates were determined for each strain as described. Results are shown grouped according to species. Each symbol represents one strain. Black colour indicates that mutants were detected and grey colour indicates that mutants did not occur. In the latter case, the limit of detection is given, which was calculated for each mutant-negative isolate by assuming that one single mutant was found (i.e. it is dependent on the number of parallel cultures and the bacterial count). In addition, the number (n) of tested isolates per species is given, and the mean mutation rates, which were calculated per species based on all calculated mutation rates and limits of detection are depicted as horizontal bars and also at the top of the figure for each organism. Mutation rates differed markedly between species. In summary, ampC-derepressed mutants occurred with mean mutation rates of 3 × 10−8 for E. cloacae complex, E. aerogenes, C. freundii complex and H. alvei, but only 2 × 10−9 for S. liquefaciens, 6 × 10−10 for Providencia spp., 2 × 10−10 for S. marcescens and 5 × 10−11 for M. morganii. Thus, compared with E. cloacae complex, E. aerogenes, C. freundii complex and H. alvei, mean mutation rates were 15-fold lower for S. liquefaciens, 50- to 150-fold lower for Providencia spp. and S. marcescens, and 600-fold lower for M. morganii. The difference between S. marcescens and S. liquefaciens indicates that identification to the species level is important because mutation rates for ampC derepression vary not only at the genus level, but also at the species level. Whereas species determination is clearly useful to predict the mutation rate, we could not identify a correlation between the β-lactam MICs of the WT strain and the mutation rate (data not shown). We performed disc diffusion and boronic acid inhibition tests to prove ampC derepression in the observed mutants. Interestingly, the in vitro antimicrobial susceptibility patterns of the mutants differed between species (Figure 2). For instance, susceptibility to ceftazidime was conserved in most Serratia mutants, and susceptibility to piperacillin/tazobactam was conserved in most M. morganii and P. stuartii mutants and half of the Serratia mutants. Figure 2. View largeDownload slide Antimicrobial susceptibility patterns of the ampC-derepressed mutants. Seven mutants were isolated from different parallel cultures for each WT strain (fewer mutants in case of low mutation rates). For each of these mutants, antimicrobial susceptibility testing was done by disc diffusion, and the boronic acid inhibition test was performed to prove ampC derepression. All mutants showed derepressed β-lactam resistance patterns (i.e. non-susceptibility to at least one oxyimino-cephalosporin) as well as zone diameter increases in the presence of APB in the disc diffusion test. For each species and the antibiotics piperacillin/tazobactam (TZP), cefotaxime (CTX) and ceftazidime (CAZ), we show the percentages of mutants that tested susceptible (black) and intermediate (grey), according to EUCAST breakpoints.16 In addition, the number of tested mutants (n) is given. Results have to be interpreted with caution for M. morganii as we detected only three mutants. Figure 2. View largeDownload slide Antimicrobial susceptibility patterns of the ampC-derepressed mutants. Seven mutants were isolated from different parallel cultures for each WT strain (fewer mutants in case of low mutation rates). For each of these mutants, antimicrobial susceptibility testing was done by disc diffusion, and the boronic acid inhibition test was performed to prove ampC derepression. All mutants showed derepressed β-lactam resistance patterns (i.e. non-susceptibility to at least one oxyimino-cephalosporin) as well as zone diameter increases in the presence of APB in the disc diffusion test. For each species and the antibiotics piperacillin/tazobactam (TZP), cefotaxime (CTX) and ceftazidime (CAZ), we show the percentages of mutants that tested susceptible (black) and intermediate (grey), according to EUCAST breakpoints.16 In addition, the number of tested mutants (n) is given. Results have to be interpreted with caution for M. morganii as we detected only three mutants. Discussion Our study was conducted to provide detailed data on the mutation rates for ampC derepression for Enterobacterales with inducible AmpC, including a large number of isolates and different species. Previous publications, which analysed AmpC regulation mechanisms and mutations leading to ampC derepression, merely described mutant frequencies of 10−6 to 10−7, based on only a small number of isolates (two to five strains) and limited to Enterobacter spp.8,9,11 or Citrobacter spp.7,10,12 However, mutant frequencies are much less accurate and reproducible than mutation rates,15 and for H. alvei, Providencia spp., Serratia spp. and M. morganii, even data on mutant frequencies are lacking. Hence, a study calculating species-specific mutation rates using Luria–Delbrück fluctuation analyses was needed. We found that mutation rates were similarly high for E. cloacae complex, E. aerogenes, C. freundii complex and H. alvei isolates, with a mean mutation rate of 3 × 10−8. In contrast, mutation rates were markedly lower for the other Enterobacterales with inducible AmpC, namely 2 × 10−9 for S. liquefaciens, 6 × 10−10 for Providencia spp., 2 × 10−10 for S. marcescens and 5 × 10−11 for M. morganii. Because the calculated mean mutation rates rely in part on the limits of detection, the true means may be overestimated in the three last-mentioned genera, for which the percentage of mutant-negative isolates was high in our study. As the limit of detection depends on the number of parallel cultures used, re-testing of mutant-negative isolates with a larger number of parallel cultures might have led to more precise and possibly lower species-specific mutation rates. However, because fluctuation assays are associated with a very high expenditure of time and effort, we abstained from increasing the number of parallel cultures beyond n = 32, with the exception of M. morganii, where we used 96 cultures. Furthermore, we cannot exclude that choice and concentration of the antibiotic used for mutant selection influences results, although this was not apparent in preliminary experiments conducted during protocol establishment. Our results are of utmost importance for clinical microbiology laboratories and infectious disease specialists because the species-specific mutation rates for ampC derepression allow estimation of the extent to which resistant mutants are present in infections caused by Enterobacterales with inducible AmpC, and this correlates with the risk of treatment failure: if the mutation rate is low, the outgrowth of resistant mutants during treatment will be slower because of a lower number of pre-existing mutants (i.e. lower ‘inoculum’). In this case, the use of, e.g. oxyimino-cephalosporins may be safe. In agreement with the present EUCAST expert rules,3 our data support discouraging treatment with oxyimino-cephalosporins in infections by E. cloacae complex, E. aerogenes and C. freundii complex, and we suggest including H. alvei in the rule because of similarly high mutation rates. In contrast, the risk of selecting mutants is considerably lower in WT Providencia spp. and Serratia spp., and may be negligible in M. morganii because of the markedly lower mutation rates. Thus, for the latter genera, it may be admissible to report susceptible results with addition of the comment that emergence of resistance may occur. Of note, the decision to use oxyimino-cephalosporins should consider the source and the severity of the infection because the probability of pre-existing mutants (which correlates with the probability of treatment failure) will depend highly on the number of bacteria at the site of infection. Moreover, when using oxyimino-cephalosporins, it is certainly prudent to assess treatment response and monitor for signs of breakthrough infection. Clinical data on the emergence of resistance in patients with infections caused by WT Enterobacterales with inducible AmpC are scarce. For WT Enterobacter infections, emergence of resistance during broad-spectrum cephalosporin therapy was reported in 6/31 (19%),5 31/161 (19%)6 and 10/121 patients (8%),4 but for WT C. freundii complex, S. marcescens and M. morganii emergence of resistance was rarely observed, namely in 1/39 (3%), 0/37 (0%) and 0/21 patients (0%), respectively.4 To date, it has been concluded that the risk of emergence of resistance is substantial in Enterobacter infections, but less so for other Enterobacterales with inducible AmpC.4,19 Thus, in reviews and comments on treatment of bacteria with inducible AmpC, the authors repeatedly recommended avoiding broad-spectrum cephalosporins for treatment of Enterobacter infections, at least in serious infections.19–21 Of note, the conclusions drawn from the aforementioned studies might be weakened by the small number of patients involved and the observational design. Thus, our data do not contradict those observations, because the risk might have been misjudged for C. freundii complex until now owing to an insufficient number of cases included, and because H. alvei has not been included at all. Whereas we calculated the mutation rates for ampC derepression for various isolates, it is beyond the scope of our study to elucidate the underlying mutations. Concordantly, from the clinical point of view, it is important to know at what rate mutations leading to ampC derepression occur, but not which specific genetic alteration has taken place in each individual case. However, the interested reader is referred to a number of excellent studies that elaborate on the genetic background, the AmpC induction mechanism and the mutational mechanisms.1,2 It is known that various mutations may affect AmpC activity, including AmpC constitutive hyperproduction or AmpC hyperinducibility (mediated by alterations of AmpD or AmpR), enhanced AmpC activity without strong upregulation of the ampC gene (mediated by alterations of AmpC or post-transcriptional regulation mechanisms), and non-inducible constitutive low-level production (mediated by alterations of AmpR, AmpG or NagZ).2,22–24 For E. cloacae complex and C. freundii complex, previous studies found that ampC derepression mainly results from AmpD alterations and, more rarely, from AmpR alterations, although concomitant mutations of other genes occur frequently.22,25–28 Similar studies in H. alvei, Providencia spp., Serratia spp. and M. morganii are lacking. However, increasing evidence points towards genus-specific variations in the AmpC system. For instance, various AmpC enzymes have been described,29 and the genetic environment of the ampC gene differs between species.30 In S. marcescens, AmpC production is additionally influenced by a 5′-UTR stem–loop structure affecting transcript stability.31 Furthermore, responses to induction by β-lactams have been described to depend on the bacterial genus and on the drug.2,17,22 These variations and our observation of species-specific differences in both mutation rates and antibiograms of the mutants indicate that the genetic background and, thus, types and locations of mutations may vary between genera/species. Interestingly, in four E. cloacae complex, one C. freundii complex and one H. alvei isolate we could not detect any mutants. Because these strains showed a negative AmpC induction test and in vitro susceptibility to cefoxitin, it is possible that they resembled WT strains in routine antimicrobial susceptibility testing, but actually possess a non-inducible or defective AmpC. Even though such isolates would probably be amenable to treatment with many β-lactams, it is difficult to recognize them in a clinical routine laboratory setting because phenotypic tests for inducibility are normally not performed. Furthermore, in vitro susceptibility to cefoxitin is not a reliable indicator for a non-inducible or defective AmpC because in this study all H. alvei, S. liquefaciens and Providencia spp. WT isolates were susceptible to cefoxitin, which has been acknowledged before for H. alvei.30 Moreover, a few Enterobacter spp., C. freundii complex and S. marcescens WT isolates were susceptible to cefoxitin despite high mutation rates. Notably, another C. freundii complex isolate that also showed a negative AmpC induction test still yielded mutants, albeit at a low mutation rate of 6 × 10−10, indicating that occurrence of mutants cannot be excluded in non-inducible strains. A possible explanation could be an ampR mutation resulting in an AmpR that is permanently in its activator conformation and, thus, activates ampC transcription even in non-inducible strains, in which 1,6-anhydro-N-acetylmuramic acid oligopeptides are not produced because of, for instance, a defective AmpG. The lower mutation rate is then to be expected because the more common ampD mutations are of no consequence in the absence of 1,6-anhydro-N-acetylmuramic acid oligopeptides. The antibiograms of the mutants differed between species, which might be explained by species-specific differences in AmpC types29 and mutational mechanisms. In our study, the ampC-derepressed isolates were almost never susceptible to oxyimino-cephalosporins, with the exception of Serratia spp. and ceftazidime, and zones of inhibition were smaller in resistant Enterobacter, Citrobacter and Hafnia isolates than in resistant Serratia, Providencia and Morganella isolates, which are both in line with previous reports.17,32,In vitro susceptibility rates were considerably higher for piperacillin/tazobactam, especially for M. morganii (100%), P. stuartii (96%) and Serratia spp. (55%). The clinical implication of this finding might be controversial. Theoretically, tazobactam is not an effective inhibitor of AmpC and even acts as a weak inducer.1 However, it sufficiently inhibits the AmpC produced by M. morganii, and seems to have some ability to potentiate the activity of piperacillin against ampC-derepressed mutants also in other species.33 Moreover, piperacillin/tazobactam selects resistant mutants less effectively than oxyimino-cephalosporins in in vitro and animal models,34,35 which is supported by some clinical observations4,6 but contradicted by others.36 One review on treatment of bacteria with inducible AmpC concluded that there is a moderate level of evidence for clinical efficacy of piperacillin/tazobactam and that routine suppression of laboratory susceptibility data may not be justified, especially as treatment alternatives have a less favourable side effect profile (fluoroquinolones, aminoglycosides, trimethoprim/sulfamethoxazole) or should be spared because of their extremely broad spectrum (carbapenems).19 Interestingly, two recent retrospective analyses concluded that piperacillin/tazobactam treatment of Enterobacter spp., Citrobacter spp. and Serratia spp. bacteraemia was not significantly associated with treatment failure.37,38 However, there was a clear trend towards higher 30 day mortality (15% for piperacillin/tazobactam compared with 7% for carbapenems and cefepime in the matched cohort)37 and higher risk of persistent bacteraemia (20% for piperacillin/tazobactam compared with 10% for carbapenems and cefepime in the matched cohort,37 and 29% for piperacillin/tazobactam compared with 18% for other agents38) which might have failed to reach statistical significance because of the limited number of cases. Based on our data, we think that clinicians should exercise caution in using piperacillin/tazobactam for infections caused by Enterobacter spp., C. freundii complex and H. alvei because of both high mutation rates and comparably low susceptibility rates in the mutants (6%, 29% and 1%, respectively), especially when treating infections with high bacterial loads (because this increases the probability of pre-existing mutants). On the other hand, the use of piperacillin/tazobactam might be justified for infections caused by Providencia spp., Serratia spp. and M. morganii because of lower mutation rates. Yet, as discussed above for the use of oxyimino-cephalosporins, caution might be advisable in case of severe infections and close monitoring of the patients should be ensured. In conclusion, we provide species-specific mutation rates for different Enterobacterales with inducible AmpC, and suggest avoiding the use of oxyimino-cephalosporins (and piperacillin/tazobactam) for infections caused by Enterobacter spp., C. freundii complex and H. alvei. In contrast, choosing oxyimino-cephalosporins and especially piperacillin/tazobactam might be reasonable for infections caused by Serratia spp., Providencia spp. and M. morganii, with close monitoring of the patients. However, because we present in vitro data only, we would like to stress that clinical studies should be conducted for future correlation with our present data. Thus, if the EUCAST expert rule is modified on the basis of the data presented in this work, a further adaptation might be required as soon as clinical outcome data are available. Acknowledgements The authors thank Suntke Gerriets, Ronja Reyes Henriquez, Rebecca Schlegel and Kira Bublitz, who conducted fluctuation assays under the supervision of the authors, and Susanne Friedrich and Brigitte Hemmerle for excellent technical assistance.  This work was presented, in part, at the 27th European Congress of Clinical Microbiology and Infectious Diseases (ECCMID), 22–25 April 2017, Vienna, Austria (Presentation number 649). Funding This study was supported by internal funding. Transparency declarations None to declare. References 1 Jacoby GA. AmpC β-lactamases. Clin Microbiol Rev  2009; 22: 161– 82. Google Scholar CrossRef Search ADS PubMed  2 Hanson ND, Sanders CC. Regulation of inducible AmpC β-lactamase expression among Enterobacteriaceae. Curr Pharm Des  1999; 5: 881– 94. Google Scholar PubMed  3 Leclercq R, Canton R, Brown DF et al.   EUCAST expert rules in antimicrobial susceptibility testing. Clin Microbiol Infect  2013; 19: 141– 60. Google Scholar CrossRef Search ADS PubMed  4 Choi SH, Lee JE, Park SJ et al.   Emergence of antibiotic resistance during therapy for infections caused by Enterobacteriaceae producing AmpC β-lactamase: implications for antibiotic use. Antimicrob Agents Chemother  2008; 52: 995– 1000. Google Scholar CrossRef Search ADS PubMed  5 Chow JW, Fine MJ, Shlaes DM et al.   Enterobacter bacteremia: clinical features and emergence of antibiotic resistance during therapy. Ann Intern Med  1991; 115: 585– 90. Google Scholar CrossRef Search ADS PubMed  6 Kaye KS, Cosgrove S, Harris A et al.   Risk factors for emergence of resistance to broad-spectrum cephalosporins among Enterobacter spp. Antimicrob Agents Chemother  2001; 45: 2628– 30. Google Scholar CrossRef Search ADS PubMed  7 Gootz TD, Jackson DB, Sherris JC. Development of resistance to cephalosporins in clinical strains of Citrobacter spp. Antimicrob Agents Chemother  1984; 25: 591– 5. Google Scholar CrossRef Search ADS PubMed  8 Gootz TD, Sanders CC, Goering RV. Resistance to cefamandole: derepression of β-lactamases by cefoxitin and mutation in Enterobacter cloacae. J Infect Dis  1982; 146: 34– 42. Google Scholar CrossRef Search ADS PubMed  9 Lampe MF, Allan BJ, Minshew BH et al.   Mutational enzymatic resistance of Enterobacter species to β-lactam antibiotics. Antimicrob Agents Chemother  1982; 21: 655– 60. Google Scholar CrossRef Search ADS PubMed  10 Lindberg F, Westman L, Normark S. Regulatory components in Citrobacter freundii ampC β-lactamase induction. Proc Natl Acad Sci USA  1985; 82: 4620– 4. Google Scholar CrossRef Search ADS PubMed  11 Seeberg AH, Tolxdorff-Neutzling RM, Wiedemann B. Chromosomal β-lactamases of Enterobacter cloacae are responsible for resistance to third-generation cephalosporins. Antimicrob Agents Chemother  1983; 23: 918– 25. Google Scholar CrossRef Search ADS PubMed  12 Stapleton P, Shannon K, Phillips I. The ability of β-lactam antibiotics to select mutants with derepressed β-lactamase synthesis from Citrobacter freundii. J Antimicrob Chemother  1995; 36: 483– 96. Google Scholar CrossRef Search ADS PubMed  13 Chan WC, Li RC, Ling JM et al.   Markedly different rates and resistance profiles exhibited by seven commonly used and newer β-lactams on the selection of resistant variants of Enterobacter cloacae. J Antimicrob Chemother  1999; 43: 55– 60. Google Scholar CrossRef Search ADS PubMed  14 Fung-Tomc JC, Gradelski E, Huczko E et al.   Differences in the resistant variants of Enterobacter cloacae selected by extended-spectrum cephalosporins. Antimicrob Agents Chemother  1996; 40: 1289– 93. Google Scholar PubMed  15 Rosche WA, Foster PL. Determining mutation rates in bacterial populations. Methods  2000; 20: 4– 17. Google Scholar CrossRef Search ADS PubMed  16 The European Committee on Antimicrobial Susceptibility Testing. Breakpoint Tables for Interpretation of MICs and Zone Diameters. Version 7.1, 2017. http://www.eucast.org. 17 Sanders CC, Sanders WEJr. Type I β-lactamases of gram-negative bacteria: interactions with β-lactam antibiotics. J Infect Dis  1986; 154: 792– 800. Google Scholar CrossRef Search ADS PubMed  18 Armitage P. The statistical theory of bacterial populations subject to mutation. J R Statist Soc B  1952; 14: 1– 40. 19 Harris PN, Ferguson JK. Antibiotic therapy for inducible AmpC β-lactamase-producing Gram-negative bacilli: what are the alternatives to carbapenems, quinolones and aminoglycosides? Int J Antimicrob Agents  2012; 40: 297– 305. Google Scholar CrossRef Search ADS PubMed  20 Livermore DM, Brown DF, Quinn JP et al.   Should third-generation cephalosporins be avoided against AmpC-inducible Enterobacteriaceae? Clin Microbiol Infect  2004; 10: 84– 5. Google Scholar CrossRef Search ADS PubMed  21 Macdougall C. Beyond susceptible and resistant, part I: treatment of infections due to Gram-negative organisms with inducible β-lactamases. J Pediatr Pharmacol Ther  2011; 16: 23– 30. Google Scholar PubMed  22 Guerin F, Isnard C, Cattoir V et al.   Complex regulation pathways of AmpC-mediated β-lactam resistance in Enterobacter cloacae complex. Antimicrob Agents Chemother  2015; 59: 7753– 61. Google Scholar CrossRef Search ADS PubMed  23 Haruta S, Nukaga M, Taniguchi K et al.   Resistance to oxyimino β-lactams due to a mutation of chromosomal β-lactamase in Citrobacter freundii. Microbiol Immunol  1998; 42: 165– 9. Google Scholar CrossRef Search ADS PubMed  24 Matsumura N, Minami S, Mitsuhashi S. Sequences of homologous β-lactamases from clinical isolates of Serratia marcescens with different substrate specificities. Antimicrob Agents Chemother  1998; 42: 176– 9. Google Scholar PubMed  25 Kaneko K, Okamoto R, Nakano R et al.   Gene mutations responsible for overexpression of AmpC β-lactamase in some clinical isolates of Enterobacter cloacae. J Clin Microbiol  2005; 43: 2955– 8. Google Scholar CrossRef Search ADS PubMed  26 Kopp U, Wiedemann B, Lindquist S et al.   Sequences of wild-type and mutant ampD genes of Citrobacter freundii and Enterobacter cloacae. Antimicrob Agents Chemother  1993; 37: 224– 8. Google Scholar CrossRef Search ADS PubMed  27 Kuga A, Okamoto R, Inoue M. ampR gene mutations that greatly increase class C β-lactamase activity in Enterobacter cloacae. Antimicrob Agents Chemother  2000; 44: 561– 7. Google Scholar CrossRef Search ADS PubMed  28 Petrosino JF, Pendleton AR, Weiner JH et al.   Chromosomal system for studying AmpC-mediated β-lactam resistance mutation in Escherichia coli. Antimicrob Agents Chemother  2002; 46: 1535– 9. Google Scholar CrossRef Search ADS PubMed  29 Philippon A, Arlet G, Jacoby GA. Plasmid-determined AmpC-type β-lactamases. Antimicrob Agents Chemother  2002; 46: 1– 11. Google Scholar CrossRef Search ADS PubMed  30 Girlich D, Naas T, Bellais S et al.   Heterogeneity of AmpC cephalosporinases of Hafnia alvei clinical isolates expressing inducible or constitutive ceftazidime resistance phenotypes. Antimicrob Agents Chemother  2000; 44: 3220– 3. Google Scholar CrossRef Search ADS PubMed  31 Mahlen SD, Morrow SS, Abdalhamid B et al.   Analyses of ampC gene expression in Serratia marcescens reveal new regulatory properties. J Antimicrob Chemother  2003; 51: 791– 802. Google Scholar CrossRef Search ADS PubMed  32 Curtis NA, Eisenstadt RL, Rudd C et al.   Inducible type I β-lactamases of gram-negative bacteria and resistance to β-lactam antibiotics. J Antimicrob Chemother  1986; 17: 51– 61. Google Scholar CrossRef Search ADS PubMed  33 Akova M, Yang Y, Livermore DM. Interactions of tazobactam and clavulanate with inducibly- and constitutively-expressed Class I β-lactamases. J Antimicrob Chemother  1990; 25: 199– 208. Google Scholar CrossRef Search ADS PubMed  34 Higashitani F, Nishida K, Hyodo A et al.   Effects of tazobactam on the frequency of the emergence of resistant strains from Enterobacter cloacae, Citrobacter freundii, and Proteus vulgaris (β-lactamase derepressed mutants). J Antibiot (Tokyo)  1995; 48: 1027– 33. Google Scholar CrossRef Search ADS PubMed  35 Stearne LE, van Boxtel D, Lemmens N et al.   Comparative study of the effects of ceftizoxime, piperacillin, and piperacillin-tazobactam concentrations on antibacterial activity and selection of antibiotic-resistant mutants of Enterobacter cloacae and Bacteroides fragilis in vitro and in vivo in mixed-infection abscesses. Antimicrob Agents Chemother  2004; 48: 1688– 98. Google Scholar CrossRef Search ADS PubMed  36 Schwaber MJ, Graham CS, Sands BE et al.   Treatment with a broad-spectrum cephalosporin versus piperacillin-tazobactam and the risk for isolation of broad-spectrum cephalosporin-resistant Enterobacter species. Antimicrob Agents Chemother  2003; 47: 1882– 6. Google Scholar CrossRef Search ADS PubMed  37 Cheng L, Nelson BC, Mehta M et al.   Piperacillin-tazobactam versus other antibacterial agents for treatment of bloodstream infections due to AmpC β-lactamase-producing Enterobacteriaceae. Antimicrob Agents Chemother  2017; 61: pii=e00276-17. 38 Harris PNA, Peri AM, Pelecanos AM et al.   Risk factors for relapse or persistence of bacteraemia caused by Enterobacter spp.: a case-control study. Antimicrob Resist Infect Control  2017; 6: 14. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For permissions, please email: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Antimicrobial Chemotherapy Oxford University Press

Species-specific mutation rates for ampC derepression in Enterobacterales with chromosomally encoded inducible AmpC β-lactamase

Loading next page...
 
/lp/ou_press/species-specific-mutation-rates-for-ampc-derepression-in-EeVodwuBmy
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For permissions, please email: journals.permissions@oup.com.
ISSN
0305-7453
eISSN
1460-2091
D.O.I.
10.1093/jac/dky084
Publisher site
See Article on Publisher Site

Abstract

Abstract Background AmpC β-lactamases are encoded on the chromosomes of certain Enterobacterales and lead to clinical resistance to various β-lactams in case of high-level expression. In WT bacteria with inducible AmpC, the expression is low, but selection of stably ampC-derepressed mutants may occur during β-lactam therapy. Thus, for Enterobacter spp., Citrobacter freundii complex, Serratia spp. and Morganella morganii that test susceptible in vitro to oxyimino-cephalosporins, the EUCAST expert rules recommend suppressing susceptibility testing results for these agents or noting that their use in monotherapy should be discouraged, owing to the risk of selecting resistance. However, clinical observations suggest that emergence of resistance is not equally common in all species with inducible AmpC. Objectives To determine species-specific mutation rates, which are more accurate and reproducible than previously described mutant frequencies, for ampC derepression in Enterobacterales with inducible AmpC. Methods Mutation rates were determined using a protocol based on Luria–Delbrück fluctuation analyses. Overall, 237 isolates were analysed. Results Mutation rates were high in Enterobacter cloacae complex, Enterobacter aerogenes, C. freundii complex and Hafnia alvei isolates, with a mean mutation rate of 3 × 10−8. In contrast, mean mutation rates were considerably lower in Providencia spp., Serratia spp. and especially M. morganii isolates. Furthermore, we observed species-specific variations in the resistance patterns of ampC-derepressed mutants. Conclusions Our data might help to predict the risk of treatment failure with oxyimino-cephalosporins in infections by different Enterobacterales with inducible AmpC. Moreover, we make a proposal for optimization of the current EUCAST expert rule. Introduction Inducible AmpC β-lactamases are clinically important enzymes encoded on the chromosomes of certain Enterobacterales, mainly including Enterobacter cloacae complex, Enterobacter aerogenes, Citrobacter freundii complex, Hafnia alvei, Providencia spp., Serratia spp. and Morganella morganii.1 They hydrolyse penicillins, cephalosporins and monobactams. In WT bacteria, AmpC expression is low but inducible (via a complex regulation mechanism) in response to β-lactam exposure, and mediates clinical resistance to β-lactams that are strong inducers (cephamycins such as cefoxitin, first/second-generation cephalosporins such as cefuroxime, and aminopenicillins±β-lactamase inhibitors) but not to β-lactams that are weak inducers (e.g. oxyimino-cephalosporins).1 However, mutations may occur spontaneously, resulting in stable derepression of the ampC gene.1,2 The subsequent constant high-level expression of AmpC leads to clinical resistance to all β-lactams apart from carbapenems and sometimes cefepime.1 In infections by WT bacteria with inducible AmpC, therapy with, e.g. oxyimino-cephalosporins can select for pre-existing stably ampC-derepressed mutants, resulting in breakthrough infection and treatment failure. Thus, for Enterobacter spp., C. freundii complex, Serratia spp. and M. morganii testing as susceptible in vitro to oxyimino-cephalosporins, the EUCAST expert rules recommend suppressing susceptibility testing results for these agents or noting that their use in monotherapy should be discouraged, owing to the risk of selecting resistance.3 Of note, clinical observations suggest that emergence of resistance is more common in Enterobacter infections (8%–19%) than in infections by other species (e.g. 3% in Citrobacter infections),4–6 but data are scarce. Knowing exact species-specific mutation rates for ampC derepression would help to predict the risk of treatment failure with oxyimino-cephalosporins in infections caused by Enterobacterales with inducible AmpC because they would allow estimation of the extent to which spontaneously arising mutants are present. To the best of our knowledge, species-specific mutation rates for ampC derepression have not been described for Enterobacterales with inducible AmpC. In previous studies, the authors merely determined mutant frequencies by simply counting the number of mutant bacteria that had grown on selective media in relation to the number of plated bacteria.7–12 Others utilized serial passaging on selective media to compare different antibiotics for their resistance selection potential.13,14 However, the results gained with both methods are inadequate indicators of spontaneous mutation rates. Luria–Delbrück fluctuation analyses are the correct method for determining mutation rates, as outlined by Rosche and Foster.15 Thus, in the present study, we performed Luria–Delbrück fluctuation analyses, which have not been applied to AmpC mutational resistance studies before, in order to provide species-specific mutation rates for ampC derepression. As Rosche and Foster15 clearly and extensively explain the theoretical basis and the methodological details, we recommend reading their paper for detailed background information on the method. Briefly, mutant frequencies (i.e. the proportion of cells in a population that are mutants) are inherently inaccurate because the population of mutants is composed of clones, each of which arose from a cell that sustained a mutation. The size of a given mutant clone (i.e. the number of its individual bacteria) will depend on when during the growth of the population the mutation occurred. Consequently, considering that bacteria grow exponentially and mutations arise stochastically, the fraction of mutants is largely influenced by the timepoints at which the mutations occurred and not only by the number of mutations (i.e. the mutation rate). Hence, even if results from replicate cultures are averaged, mutant frequencies will not be a reliable and reproducible measure of the mutation rate. The distribution of mutant numbers among replicate cultures is usually referred to as the Luria–Delbrück distribution. With the help of this theoretical distribution, the mutation rate can be mathematically estimated using fluctuation analyses. A fluctuation test begins with the inoculation of a small number of cells into a large number of parallel cultures. The cultures are allowed to grow to saturation in non-selective media, and then each culture is plated on selective media for detection of the mutants that have occurred. The observed distribution of the numbers of mutants among the parallel cultures is then used to calculate the probable number of mutations per culture that gave rise to that distribution. Finally, the number of mutations is converted to the mutation rate by dividing it by some function of the total number of cells at risk. Materials and methods Specimens Randomly selected non-duplicate E. cloacae complex, E. aerogenes, C. freundii complex, H. alvei, Providencia rettgeri, Providencia stuartii, Serratia marcescens, Serratialiquefaciens and M. morganii isolates were collected between January 2016 and July 2017 during routine diagnostics in our affiliated clinical microbiology laboratory, provided that they tested as susceptible in vitro to oxyimino-cephalosporins (i.e. MIC ≤1 mg/L for cefotaxime and ceftazidime). For improved readability, these isolates are subsequently referred to as WT strains, although we did not consider whether or not resistance to other antibiotics was present. Species identification was achieved by MALDI-TOF (Microflex LT system, Bruker Daltonics; MBT Compass 4.1.31 software), and antimicrobial susceptibility testing was done according to EUCAST recommendations,16 mainly using Vitek®2 (bioMérieux). Fluctuation assay The fluctuation assay protocol was established according to the instructions of Rosche and Foster.15 For each WT isolate, multiple parallel cultures, each comprising 100 μL of a 105 cfu/mL organism suspension in Mueller–Hinton Broth (Oxoid), were incubated overnight at 37°C. The number of parallel cultures is a critical factor: if the mutation rate is very low, no mutants will arise in the parallel cultures that can be detected during the subsequent selection step, unless an extremely large number of cultures is used. Yet the occurrence of at least one mutant in at least one culture is the precondition that a mutation rate can be calculated. As a consequence, the minimal mutation rate that can be determined (i.e. the limit of detection) is dependent on the number of parallel cultures and the bacterial count per culture. We adjusted the number of parallel cultures dependent on the species, as evaluated in preliminary experiments. For Enterobacter spp., C. freundii complex and H. alvei, 16 cultures were sufficient to gain mutants in almost all WT isolates (otherwise, the experiment was repeated with 32 cultures to lower the limit of detection). For Providencia spp. and Serratia spp., we used 32 cultures, which enabled us to find mutants in a considerable number of isolates. For M. morganii, we were not able to detect any mutants using 32 cultures and, thus, extended experiments to 96 cultures. Two cultures were used to determine bacterial counts in duplicate by plating serial dilutions on non-selective media, i.e. each dilution was mixed with 8 mL of Mueller–Hinton Soft Agar (Oxoid) and then dispersed in a culture plate. The remaining cultures were plated on selective media, i.e. each entire culture was mixed with 8 mL of Mueller–Hinton Soft Agar containing cefotaxime (8 mg/L; Oxoid) and then dispersed in a culture plate. Cefotaxime is, as a weak inducer of AmpC, suitable for selection of ampC-derepressed mutants.1,12,17 Culture plates were incubated overnight at 37°C, then the colonies in/on the soft agar were counted for determination of bacterial counts and mutant numbers per parallel culture. As AmpC inducibility by β-lactams varies between genera,17 in a few strains background growth (induced WT) complicated mutant counting. Then, the experiment was repeated with another weak inducer, namely piperacillin (32 mg/L; Oxoid),1,12,17 to confirm results. In the re-tested strains, using piperacillin reduced background growth, which enabled us to gain reliable mutant counts. However, we could not generally replace cefotaxime by piperacillin for mutant selection in this study because of isolated resistance to piperacillin in some WT strains, possibly because of an additional narrow-spectrum β-lactamase such as TEM or SHV. Cefotaxime and piperacillin concentrations were chosen considering the MIC distributions published on the EUCAST web site. Based on the distribution of mutant numbers in the parallel cultures, the probable number of mutations was estimated using the MSS maximum likelihood method. Then, mutation rates per cell per generation were calculated from the number of mutations and the bacterial count, applying the In(2) correction as suggested by Armitage.18 The formulas used for the calculations are described in detail in the publication by Rosche and Foster.15 For isolates that yielded no mutants (subsequently referred to as mutant-negative isolates, although mutants might have arisen in case of a larger number of parallel cultures), a mutation rate could not be calculated. Thus, we calculated the limit of detection instead, by assuming that exactly one mutant was found in one parallel culture. Phenotypic characterization of WT strains and mutants Disc diffusion using Oxoid discs was done for WT strains and mutant colonies according to EUCAST recommendations.16 In addition, for the WT strains, we checked for presence of an inducible AmpC by a disc induction test, placing strong inducers (cefoxitin and imipenem) adjacent to weak inducers (cefotaxime, ceftazidime and piperacillin). To confirm ampC derepression as the underlying resistance mechanism, mutants were subjected to the boronic acid inhibition test as boronic acid is a well-known inhibitor of AmpC.1 The test was performed by the standard disc diffusion method, using cefotaxime discs alone and in combination with 300 μg of 3-aminophenylboronic acid (APB; ACROS Organics). We tested seven representative mutant colonies from different parallel cultures, or all mutants if fewer than seven cultures yielded mutants. Results We established a fluctuation assay protocol for calculation of mutation rates for ampC derepression for Enterobacterales with inducible AmpC. The number of WT isolates analysed per species, the mutation rate for each isolate and mean mutation rates per species are shown in Figure 1. With the selected number of parallel cultures, we observed mutants in 126/132 (95%) Enterobacter spp., C. freundii complex and H. alvei WT isolates (of note, in the six mutant-negative isolates, the AmpC was not inducible in the disc induction test). In the other species, mutants occurred in only 33/80 (41%) Providencia spp. and Serratia spp. isolates, and in only 3/25 (12%) M. morganii isolates despite the higher number of parallel cultures. Figure 1. View largeDownload slide Species-specific mutation rates. Mutation rates were determined for each strain as described. Results are shown grouped according to species. Each symbol represents one strain. Black colour indicates that mutants were detected and grey colour indicates that mutants did not occur. In the latter case, the limit of detection is given, which was calculated for each mutant-negative isolate by assuming that one single mutant was found (i.e. it is dependent on the number of parallel cultures and the bacterial count). In addition, the number (n) of tested isolates per species is given, and the mean mutation rates, which were calculated per species based on all calculated mutation rates and limits of detection are depicted as horizontal bars and also at the top of the figure for each organism. Figure 1. View largeDownload slide Species-specific mutation rates. Mutation rates were determined for each strain as described. Results are shown grouped according to species. Each symbol represents one strain. Black colour indicates that mutants were detected and grey colour indicates that mutants did not occur. In the latter case, the limit of detection is given, which was calculated for each mutant-negative isolate by assuming that one single mutant was found (i.e. it is dependent on the number of parallel cultures and the bacterial count). In addition, the number (n) of tested isolates per species is given, and the mean mutation rates, which were calculated per species based on all calculated mutation rates and limits of detection are depicted as horizontal bars and also at the top of the figure for each organism. Mutation rates differed markedly between species. In summary, ampC-derepressed mutants occurred with mean mutation rates of 3 × 10−8 for E. cloacae complex, E. aerogenes, C. freundii complex and H. alvei, but only 2 × 10−9 for S. liquefaciens, 6 × 10−10 for Providencia spp., 2 × 10−10 for S. marcescens and 5 × 10−11 for M. morganii. Thus, compared with E. cloacae complex, E. aerogenes, C. freundii complex and H. alvei, mean mutation rates were 15-fold lower for S. liquefaciens, 50- to 150-fold lower for Providencia spp. and S. marcescens, and 600-fold lower for M. morganii. The difference between S. marcescens and S. liquefaciens indicates that identification to the species level is important because mutation rates for ampC derepression vary not only at the genus level, but also at the species level. Whereas species determination is clearly useful to predict the mutation rate, we could not identify a correlation between the β-lactam MICs of the WT strain and the mutation rate (data not shown). We performed disc diffusion and boronic acid inhibition tests to prove ampC derepression in the observed mutants. Interestingly, the in vitro antimicrobial susceptibility patterns of the mutants differed between species (Figure 2). For instance, susceptibility to ceftazidime was conserved in most Serratia mutants, and susceptibility to piperacillin/tazobactam was conserved in most M. morganii and P. stuartii mutants and half of the Serratia mutants. Figure 2. View largeDownload slide Antimicrobial susceptibility patterns of the ampC-derepressed mutants. Seven mutants were isolated from different parallel cultures for each WT strain (fewer mutants in case of low mutation rates). For each of these mutants, antimicrobial susceptibility testing was done by disc diffusion, and the boronic acid inhibition test was performed to prove ampC derepression. All mutants showed derepressed β-lactam resistance patterns (i.e. non-susceptibility to at least one oxyimino-cephalosporin) as well as zone diameter increases in the presence of APB in the disc diffusion test. For each species and the antibiotics piperacillin/tazobactam (TZP), cefotaxime (CTX) and ceftazidime (CAZ), we show the percentages of mutants that tested susceptible (black) and intermediate (grey), according to EUCAST breakpoints.16 In addition, the number of tested mutants (n) is given. Results have to be interpreted with caution for M. morganii as we detected only three mutants. Figure 2. View largeDownload slide Antimicrobial susceptibility patterns of the ampC-derepressed mutants. Seven mutants were isolated from different parallel cultures for each WT strain (fewer mutants in case of low mutation rates). For each of these mutants, antimicrobial susceptibility testing was done by disc diffusion, and the boronic acid inhibition test was performed to prove ampC derepression. All mutants showed derepressed β-lactam resistance patterns (i.e. non-susceptibility to at least one oxyimino-cephalosporin) as well as zone diameter increases in the presence of APB in the disc diffusion test. For each species and the antibiotics piperacillin/tazobactam (TZP), cefotaxime (CTX) and ceftazidime (CAZ), we show the percentages of mutants that tested susceptible (black) and intermediate (grey), according to EUCAST breakpoints.16 In addition, the number of tested mutants (n) is given. Results have to be interpreted with caution for M. morganii as we detected only three mutants. Discussion Our study was conducted to provide detailed data on the mutation rates for ampC derepression for Enterobacterales with inducible AmpC, including a large number of isolates and different species. Previous publications, which analysed AmpC regulation mechanisms and mutations leading to ampC derepression, merely described mutant frequencies of 10−6 to 10−7, based on only a small number of isolates (two to five strains) and limited to Enterobacter spp.8,9,11 or Citrobacter spp.7,10,12 However, mutant frequencies are much less accurate and reproducible than mutation rates,15 and for H. alvei, Providencia spp., Serratia spp. and M. morganii, even data on mutant frequencies are lacking. Hence, a study calculating species-specific mutation rates using Luria–Delbrück fluctuation analyses was needed. We found that mutation rates were similarly high for E. cloacae complex, E. aerogenes, C. freundii complex and H. alvei isolates, with a mean mutation rate of 3 × 10−8. In contrast, mutation rates were markedly lower for the other Enterobacterales with inducible AmpC, namely 2 × 10−9 for S. liquefaciens, 6 × 10−10 for Providencia spp., 2 × 10−10 for S. marcescens and 5 × 10−11 for M. morganii. Because the calculated mean mutation rates rely in part on the limits of detection, the true means may be overestimated in the three last-mentioned genera, for which the percentage of mutant-negative isolates was high in our study. As the limit of detection depends on the number of parallel cultures used, re-testing of mutant-negative isolates with a larger number of parallel cultures might have led to more precise and possibly lower species-specific mutation rates. However, because fluctuation assays are associated with a very high expenditure of time and effort, we abstained from increasing the number of parallel cultures beyond n = 32, with the exception of M. morganii, where we used 96 cultures. Furthermore, we cannot exclude that choice and concentration of the antibiotic used for mutant selection influences results, although this was not apparent in preliminary experiments conducted during protocol establishment. Our results are of utmost importance for clinical microbiology laboratories and infectious disease specialists because the species-specific mutation rates for ampC derepression allow estimation of the extent to which resistant mutants are present in infections caused by Enterobacterales with inducible AmpC, and this correlates with the risk of treatment failure: if the mutation rate is low, the outgrowth of resistant mutants during treatment will be slower because of a lower number of pre-existing mutants (i.e. lower ‘inoculum’). In this case, the use of, e.g. oxyimino-cephalosporins may be safe. In agreement with the present EUCAST expert rules,3 our data support discouraging treatment with oxyimino-cephalosporins in infections by E. cloacae complex, E. aerogenes and C. freundii complex, and we suggest including H. alvei in the rule because of similarly high mutation rates. In contrast, the risk of selecting mutants is considerably lower in WT Providencia spp. and Serratia spp., and may be negligible in M. morganii because of the markedly lower mutation rates. Thus, for the latter genera, it may be admissible to report susceptible results with addition of the comment that emergence of resistance may occur. Of note, the decision to use oxyimino-cephalosporins should consider the source and the severity of the infection because the probability of pre-existing mutants (which correlates with the probability of treatment failure) will depend highly on the number of bacteria at the site of infection. Moreover, when using oxyimino-cephalosporins, it is certainly prudent to assess treatment response and monitor for signs of breakthrough infection. Clinical data on the emergence of resistance in patients with infections caused by WT Enterobacterales with inducible AmpC are scarce. For WT Enterobacter infections, emergence of resistance during broad-spectrum cephalosporin therapy was reported in 6/31 (19%),5 31/161 (19%)6 and 10/121 patients (8%),4 but for WT C. freundii complex, S. marcescens and M. morganii emergence of resistance was rarely observed, namely in 1/39 (3%), 0/37 (0%) and 0/21 patients (0%), respectively.4 To date, it has been concluded that the risk of emergence of resistance is substantial in Enterobacter infections, but less so for other Enterobacterales with inducible AmpC.4,19 Thus, in reviews and comments on treatment of bacteria with inducible AmpC, the authors repeatedly recommended avoiding broad-spectrum cephalosporins for treatment of Enterobacter infections, at least in serious infections.19–21 Of note, the conclusions drawn from the aforementioned studies might be weakened by the small number of patients involved and the observational design. Thus, our data do not contradict those observations, because the risk might have been misjudged for C. freundii complex until now owing to an insufficient number of cases included, and because H. alvei has not been included at all. Whereas we calculated the mutation rates for ampC derepression for various isolates, it is beyond the scope of our study to elucidate the underlying mutations. Concordantly, from the clinical point of view, it is important to know at what rate mutations leading to ampC derepression occur, but not which specific genetic alteration has taken place in each individual case. However, the interested reader is referred to a number of excellent studies that elaborate on the genetic background, the AmpC induction mechanism and the mutational mechanisms.1,2 It is known that various mutations may affect AmpC activity, including AmpC constitutive hyperproduction or AmpC hyperinducibility (mediated by alterations of AmpD or AmpR), enhanced AmpC activity without strong upregulation of the ampC gene (mediated by alterations of AmpC or post-transcriptional regulation mechanisms), and non-inducible constitutive low-level production (mediated by alterations of AmpR, AmpG or NagZ).2,22–24 For E. cloacae complex and C. freundii complex, previous studies found that ampC derepression mainly results from AmpD alterations and, more rarely, from AmpR alterations, although concomitant mutations of other genes occur frequently.22,25–28 Similar studies in H. alvei, Providencia spp., Serratia spp. and M. morganii are lacking. However, increasing evidence points towards genus-specific variations in the AmpC system. For instance, various AmpC enzymes have been described,29 and the genetic environment of the ampC gene differs between species.30 In S. marcescens, AmpC production is additionally influenced by a 5′-UTR stem–loop structure affecting transcript stability.31 Furthermore, responses to induction by β-lactams have been described to depend on the bacterial genus and on the drug.2,17,22 These variations and our observation of species-specific differences in both mutation rates and antibiograms of the mutants indicate that the genetic background and, thus, types and locations of mutations may vary between genera/species. Interestingly, in four E. cloacae complex, one C. freundii complex and one H. alvei isolate we could not detect any mutants. Because these strains showed a negative AmpC induction test and in vitro susceptibility to cefoxitin, it is possible that they resembled WT strains in routine antimicrobial susceptibility testing, but actually possess a non-inducible or defective AmpC. Even though such isolates would probably be amenable to treatment with many β-lactams, it is difficult to recognize them in a clinical routine laboratory setting because phenotypic tests for inducibility are normally not performed. Furthermore, in vitro susceptibility to cefoxitin is not a reliable indicator for a non-inducible or defective AmpC because in this study all H. alvei, S. liquefaciens and Providencia spp. WT isolates were susceptible to cefoxitin, which has been acknowledged before for H. alvei.30 Moreover, a few Enterobacter spp., C. freundii complex and S. marcescens WT isolates were susceptible to cefoxitin despite high mutation rates. Notably, another C. freundii complex isolate that also showed a negative AmpC induction test still yielded mutants, albeit at a low mutation rate of 6 × 10−10, indicating that occurrence of mutants cannot be excluded in non-inducible strains. A possible explanation could be an ampR mutation resulting in an AmpR that is permanently in its activator conformation and, thus, activates ampC transcription even in non-inducible strains, in which 1,6-anhydro-N-acetylmuramic acid oligopeptides are not produced because of, for instance, a defective AmpG. The lower mutation rate is then to be expected because the more common ampD mutations are of no consequence in the absence of 1,6-anhydro-N-acetylmuramic acid oligopeptides. The antibiograms of the mutants differed between species, which might be explained by species-specific differences in AmpC types29 and mutational mechanisms. In our study, the ampC-derepressed isolates were almost never susceptible to oxyimino-cephalosporins, with the exception of Serratia spp. and ceftazidime, and zones of inhibition were smaller in resistant Enterobacter, Citrobacter and Hafnia isolates than in resistant Serratia, Providencia and Morganella isolates, which are both in line with previous reports.17,32,In vitro susceptibility rates were considerably higher for piperacillin/tazobactam, especially for M. morganii (100%), P. stuartii (96%) and Serratia spp. (55%). The clinical implication of this finding might be controversial. Theoretically, tazobactam is not an effective inhibitor of AmpC and even acts as a weak inducer.1 However, it sufficiently inhibits the AmpC produced by M. morganii, and seems to have some ability to potentiate the activity of piperacillin against ampC-derepressed mutants also in other species.33 Moreover, piperacillin/tazobactam selects resistant mutants less effectively than oxyimino-cephalosporins in in vitro and animal models,34,35 which is supported by some clinical observations4,6 but contradicted by others.36 One review on treatment of bacteria with inducible AmpC concluded that there is a moderate level of evidence for clinical efficacy of piperacillin/tazobactam and that routine suppression of laboratory susceptibility data may not be justified, especially as treatment alternatives have a less favourable side effect profile (fluoroquinolones, aminoglycosides, trimethoprim/sulfamethoxazole) or should be spared because of their extremely broad spectrum (carbapenems).19 Interestingly, two recent retrospective analyses concluded that piperacillin/tazobactam treatment of Enterobacter spp., Citrobacter spp. and Serratia spp. bacteraemia was not significantly associated with treatment failure.37,38 However, there was a clear trend towards higher 30 day mortality (15% for piperacillin/tazobactam compared with 7% for carbapenems and cefepime in the matched cohort)37 and higher risk of persistent bacteraemia (20% for piperacillin/tazobactam compared with 10% for carbapenems and cefepime in the matched cohort,37 and 29% for piperacillin/tazobactam compared with 18% for other agents38) which might have failed to reach statistical significance because of the limited number of cases. Based on our data, we think that clinicians should exercise caution in using piperacillin/tazobactam for infections caused by Enterobacter spp., C. freundii complex and H. alvei because of both high mutation rates and comparably low susceptibility rates in the mutants (6%, 29% and 1%, respectively), especially when treating infections with high bacterial loads (because this increases the probability of pre-existing mutants). On the other hand, the use of piperacillin/tazobactam might be justified for infections caused by Providencia spp., Serratia spp. and M. morganii because of lower mutation rates. Yet, as discussed above for the use of oxyimino-cephalosporins, caution might be advisable in case of severe infections and close monitoring of the patients should be ensured. In conclusion, we provide species-specific mutation rates for different Enterobacterales with inducible AmpC, and suggest avoiding the use of oxyimino-cephalosporins (and piperacillin/tazobactam) for infections caused by Enterobacter spp., C. freundii complex and H. alvei. In contrast, choosing oxyimino-cephalosporins and especially piperacillin/tazobactam might be reasonable for infections caused by Serratia spp., Providencia spp. and M. morganii, with close monitoring of the patients. However, because we present in vitro data only, we would like to stress that clinical studies should be conducted for future correlation with our present data. Thus, if the EUCAST expert rule is modified on the basis of the data presented in this work, a further adaptation might be required as soon as clinical outcome data are available. Acknowledgements The authors thank Suntke Gerriets, Ronja Reyes Henriquez, Rebecca Schlegel and Kira Bublitz, who conducted fluctuation assays under the supervision of the authors, and Susanne Friedrich and Brigitte Hemmerle for excellent technical assistance.  This work was presented, in part, at the 27th European Congress of Clinical Microbiology and Infectious Diseases (ECCMID), 22–25 April 2017, Vienna, Austria (Presentation number 649). Funding This study was supported by internal funding. Transparency declarations None to declare. References 1 Jacoby GA. AmpC β-lactamases. Clin Microbiol Rev  2009; 22: 161– 82. Google Scholar CrossRef Search ADS PubMed  2 Hanson ND, Sanders CC. Regulation of inducible AmpC β-lactamase expression among Enterobacteriaceae. Curr Pharm Des  1999; 5: 881– 94. Google Scholar PubMed  3 Leclercq R, Canton R, Brown DF et al.   EUCAST expert rules in antimicrobial susceptibility testing. Clin Microbiol Infect  2013; 19: 141– 60. Google Scholar CrossRef Search ADS PubMed  4 Choi SH, Lee JE, Park SJ et al.   Emergence of antibiotic resistance during therapy for infections caused by Enterobacteriaceae producing AmpC β-lactamase: implications for antibiotic use. Antimicrob Agents Chemother  2008; 52: 995– 1000. Google Scholar CrossRef Search ADS PubMed  5 Chow JW, Fine MJ, Shlaes DM et al.   Enterobacter bacteremia: clinical features and emergence of antibiotic resistance during therapy. Ann Intern Med  1991; 115: 585– 90. Google Scholar CrossRef Search ADS PubMed  6 Kaye KS, Cosgrove S, Harris A et al.   Risk factors for emergence of resistance to broad-spectrum cephalosporins among Enterobacter spp. Antimicrob Agents Chemother  2001; 45: 2628– 30. Google Scholar CrossRef Search ADS PubMed  7 Gootz TD, Jackson DB, Sherris JC. Development of resistance to cephalosporins in clinical strains of Citrobacter spp. Antimicrob Agents Chemother  1984; 25: 591– 5. Google Scholar CrossRef Search ADS PubMed  8 Gootz TD, Sanders CC, Goering RV. Resistance to cefamandole: derepression of β-lactamases by cefoxitin and mutation in Enterobacter cloacae. J Infect Dis  1982; 146: 34– 42. Google Scholar CrossRef Search ADS PubMed  9 Lampe MF, Allan BJ, Minshew BH et al.   Mutational enzymatic resistance of Enterobacter species to β-lactam antibiotics. Antimicrob Agents Chemother  1982; 21: 655– 60. Google Scholar CrossRef Search ADS PubMed  10 Lindberg F, Westman L, Normark S. Regulatory components in Citrobacter freundii ampC β-lactamase induction. Proc Natl Acad Sci USA  1985; 82: 4620– 4. Google Scholar CrossRef Search ADS PubMed  11 Seeberg AH, Tolxdorff-Neutzling RM, Wiedemann B. Chromosomal β-lactamases of Enterobacter cloacae are responsible for resistance to third-generation cephalosporins. Antimicrob Agents Chemother  1983; 23: 918– 25. Google Scholar CrossRef Search ADS PubMed  12 Stapleton P, Shannon K, Phillips I. The ability of β-lactam antibiotics to select mutants with derepressed β-lactamase synthesis from Citrobacter freundii. J Antimicrob Chemother  1995; 36: 483– 96. Google Scholar CrossRef Search ADS PubMed  13 Chan WC, Li RC, Ling JM et al.   Markedly different rates and resistance profiles exhibited by seven commonly used and newer β-lactams on the selection of resistant variants of Enterobacter cloacae. J Antimicrob Chemother  1999; 43: 55– 60. Google Scholar CrossRef Search ADS PubMed  14 Fung-Tomc JC, Gradelski E, Huczko E et al.   Differences in the resistant variants of Enterobacter cloacae selected by extended-spectrum cephalosporins. Antimicrob Agents Chemother  1996; 40: 1289– 93. Google Scholar PubMed  15 Rosche WA, Foster PL. Determining mutation rates in bacterial populations. Methods  2000; 20: 4– 17. Google Scholar CrossRef Search ADS PubMed  16 The European Committee on Antimicrobial Susceptibility Testing. Breakpoint Tables for Interpretation of MICs and Zone Diameters. Version 7.1, 2017. http://www.eucast.org. 17 Sanders CC, Sanders WEJr. Type I β-lactamases of gram-negative bacteria: interactions with β-lactam antibiotics. J Infect Dis  1986; 154: 792– 800. Google Scholar CrossRef Search ADS PubMed  18 Armitage P. The statistical theory of bacterial populations subject to mutation. J R Statist Soc B  1952; 14: 1– 40. 19 Harris PN, Ferguson JK. Antibiotic therapy for inducible AmpC β-lactamase-producing Gram-negative bacilli: what are the alternatives to carbapenems, quinolones and aminoglycosides? Int J Antimicrob Agents  2012; 40: 297– 305. Google Scholar CrossRef Search ADS PubMed  20 Livermore DM, Brown DF, Quinn JP et al.   Should third-generation cephalosporins be avoided against AmpC-inducible Enterobacteriaceae? Clin Microbiol Infect  2004; 10: 84– 5. Google Scholar CrossRef Search ADS PubMed  21 Macdougall C. Beyond susceptible and resistant, part I: treatment of infections due to Gram-negative organisms with inducible β-lactamases. J Pediatr Pharmacol Ther  2011; 16: 23– 30. Google Scholar PubMed  22 Guerin F, Isnard C, Cattoir V et al.   Complex regulation pathways of AmpC-mediated β-lactam resistance in Enterobacter cloacae complex. Antimicrob Agents Chemother  2015; 59: 7753– 61. Google Scholar CrossRef Search ADS PubMed  23 Haruta S, Nukaga M, Taniguchi K et al.   Resistance to oxyimino β-lactams due to a mutation of chromosomal β-lactamase in Citrobacter freundii. Microbiol Immunol  1998; 42: 165– 9. Google Scholar CrossRef Search ADS PubMed  24 Matsumura N, Minami S, Mitsuhashi S. Sequences of homologous β-lactamases from clinical isolates of Serratia marcescens with different substrate specificities. Antimicrob Agents Chemother  1998; 42: 176– 9. Google Scholar PubMed  25 Kaneko K, Okamoto R, Nakano R et al.   Gene mutations responsible for overexpression of AmpC β-lactamase in some clinical isolates of Enterobacter cloacae. J Clin Microbiol  2005; 43: 2955– 8. Google Scholar CrossRef Search ADS PubMed  26 Kopp U, Wiedemann B, Lindquist S et al.   Sequences of wild-type and mutant ampD genes of Citrobacter freundii and Enterobacter cloacae. Antimicrob Agents Chemother  1993; 37: 224– 8. Google Scholar CrossRef Search ADS PubMed  27 Kuga A, Okamoto R, Inoue M. ampR gene mutations that greatly increase class C β-lactamase activity in Enterobacter cloacae. Antimicrob Agents Chemother  2000; 44: 561– 7. Google Scholar CrossRef Search ADS PubMed  28 Petrosino JF, Pendleton AR, Weiner JH et al.   Chromosomal system for studying AmpC-mediated β-lactam resistance mutation in Escherichia coli. Antimicrob Agents Chemother  2002; 46: 1535– 9. Google Scholar CrossRef Search ADS PubMed  29 Philippon A, Arlet G, Jacoby GA. Plasmid-determined AmpC-type β-lactamases. Antimicrob Agents Chemother  2002; 46: 1– 11. Google Scholar CrossRef Search ADS PubMed  30 Girlich D, Naas T, Bellais S et al.   Heterogeneity of AmpC cephalosporinases of Hafnia alvei clinical isolates expressing inducible or constitutive ceftazidime resistance phenotypes. Antimicrob Agents Chemother  2000; 44: 3220– 3. Google Scholar CrossRef Search ADS PubMed  31 Mahlen SD, Morrow SS, Abdalhamid B et al.   Analyses of ampC gene expression in Serratia marcescens reveal new regulatory properties. J Antimicrob Chemother  2003; 51: 791– 802. Google Scholar CrossRef Search ADS PubMed  32 Curtis NA, Eisenstadt RL, Rudd C et al.   Inducible type I β-lactamases of gram-negative bacteria and resistance to β-lactam antibiotics. J Antimicrob Chemother  1986; 17: 51– 61. Google Scholar CrossRef Search ADS PubMed  33 Akova M, Yang Y, Livermore DM. Interactions of tazobactam and clavulanate with inducibly- and constitutively-expressed Class I β-lactamases. J Antimicrob Chemother  1990; 25: 199– 208. Google Scholar CrossRef Search ADS PubMed  34 Higashitani F, Nishida K, Hyodo A et al.   Effects of tazobactam on the frequency of the emergence of resistant strains from Enterobacter cloacae, Citrobacter freundii, and Proteus vulgaris (β-lactamase derepressed mutants). J Antibiot (Tokyo)  1995; 48: 1027– 33. Google Scholar CrossRef Search ADS PubMed  35 Stearne LE, van Boxtel D, Lemmens N et al.   Comparative study of the effects of ceftizoxime, piperacillin, and piperacillin-tazobactam concentrations on antibacterial activity and selection of antibiotic-resistant mutants of Enterobacter cloacae and Bacteroides fragilis in vitro and in vivo in mixed-infection abscesses. Antimicrob Agents Chemother  2004; 48: 1688– 98. Google Scholar CrossRef Search ADS PubMed  36 Schwaber MJ, Graham CS, Sands BE et al.   Treatment with a broad-spectrum cephalosporin versus piperacillin-tazobactam and the risk for isolation of broad-spectrum cephalosporin-resistant Enterobacter species. Antimicrob Agents Chemother  2003; 47: 1882– 6. Google Scholar CrossRef Search ADS PubMed  37 Cheng L, Nelson BC, Mehta M et al.   Piperacillin-tazobactam versus other antibacterial agents for treatment of bloodstream infections due to AmpC β-lactamase-producing Enterobacteriaceae. Antimicrob Agents Chemother  2017; 61: pii=e00276-17. 38 Harris PNA, Peri AM, Pelecanos AM et al.   Risk factors for relapse or persistence of bacteraemia caused by Enterobacter spp.: a case-control study. Antimicrob Resist Infect Control  2017; 6: 14. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For permissions, please email: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Journal of Antimicrobial ChemotherapyOxford University Press

Published: Mar 16, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off