Experimental Prediction of the Natural Evolution of Antibiotic Resistance

Corresponding author: Barry G. Hall, Hutchison Hall, River Campus, University of Rochester, Rochester, NY 14627-0211., drbh{at}mail.rochester.edu (E-mail)

Communicating editor: H. OCHMAN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

The TEM family of ß-lactamases has evolved to confer resistance to most of the ß-lactam antibiotics, but not to cefepime. To determine whether the TEM ß-lactamases have the potential to evolve cefepime resistance, we evolved the ancestral TEM allele, TEM-1, in vitro and selected for cefepime resistance. After four rounds of mutagenesis and selection for increased cefepime resistance each of eight independent populations reached a level equivalent to clinical resistance. All eight evolved alleles increased the level of cefepime resistance by a factor of at least 32, and the best allele improved by a factor of 512. Sequencing showed that alleles contained from two to six amino acid substitutions, many of which were shared among alleles, and that the best allele contained only three substitutions.

WHEN penicillin was first shown to be an effective antibiotic in the 1940s, the expectation that infectious diseases could be controlled or even eradicated appeared to be realistic. Dreaded infections such as pneumonia and tuberculosis became curable, but just when we started to feel secure about our victory over microbes, antibiotic-resistant strains of bacteria started appearing at high frequencies (COHEN 2000 ). While it is true that antibiotics continue to cure most infections, it is increasingly difficult to identify antibiotics that can be used as effective treatments because of the diversity of resistant strains that exist. Resistance to every antibiotic in clinical use has been observed throughout the world (COHEN 2000 ), and in many cases microbial strains exhibit resistance to multiple antibiotics. In most cases, resistance to a new antibiotic arises within 3 years of the antibiotic's FDA approval date (MEDEIROS 1997 ).

Resistance genes produce enzymes that either modify or eliminate an antibiotic. Resistance genes can be spread by vertical transmission from parent to offspring or by horizontal transfer between different strains and species of bacteria. Sensitive microbes become resistant to antibiotics either through the acquisition of plasmid-borne resistance genes or through mutations that either upregulate the expression of a resistance gene or alter the binding or substrate specificity of an enzyme encoded by a resistance gene.

As resistance arises, the pharmaceutical industry races to create new antibiotics more quickly than the current antibiotics become obsolete (MCGOWAN 2001 ). Doctors and hospitals continuously revise their strategies for treating infections to limit the occurrence and spread of antibiotic resistance (COURVALIN and TRIEU-CUOT 2001 ). Developing effective strategies requires understanding how antibiotic resistance genes are most likely to evolve in response to the introduction of new antibiotics.

Because nature has already shown that one resistance gene can give rise to many phenotypically diverse descendant alleles, protein engineers have used in vitro evolution of resistance genes as models for exploring better strategies for protein engineering (STEMMER 1994 ). The resistance gene most commonly used as a model for protein engineering is the TEM-1 ß-lactamase. TEM-1 is one of the best-studied antibiotic resistance genes because it exists at high frequencies in antibiotic-resistant bacteria across the globe (MEDEIROS 1997 ; CHANAL et al. 2000 ; YAN et al. 2000 ). TEM-1 is clinically important because it confers resistance to penicillin and other ß-lactam antibiotics. ß-Lactams are often the preferred antibiotics because of their low toxicity and the broad spectrum of bacteria that they affect (LIVERMORE 1996 ). While TEM-1 has a spectrum that is limited to penicillins and early cephalosporins, it has given rise to >90 descendent alleles that confer resistance to most modern ß-lactam antibiotics (http://www.rochester.edu/College/BIO/labs/HallLab/TEM_Phylo.html and http://www.lahey.org/studies/webt.htm). At this point, there are only a handful of ß-lactam antibiotics to which the TEM ß-lactamases do not confer resistance. The TEMs are still evolving, and it will be valuable to know what their potential for new resistance is before that potential is realized.

While others have used the TEM enzyme to demonstrate the power of their in vitro evolution methods for protein engineering, we have developed an in vitro evolution method for the specific purpose of predicting how resistance genes will evolve in nature. We recently showed that we could reproduce the natural evolution of the TEM ß-lactamases that has occurred during the past 20 years. Phylogenetic analysis showed that nine amino acid substitutions have arisen multiple times in nature in response to the use of a class of ß-lactams known as the "extended spectrum cephalosporins." Because those nine substitutions were selected multiple times in nature, it is clear that they are important for resistance to extended spectrum cephalosporins (BARLOW and HALL 2002 ). Structural and biochemical analysis of those substitutions has confirmed their importance for increasing cephalosporin resistance (KNOX 1995 ). Among 10 alleles that were independently evolved in vitro by our method, we repeatedly recovered seven of the nine substitutions that have arisen multiple times in nature (BARLOW and HALL 2002 ). This means that our in vitro evolution method creates and selects the same mutations that are found in natural TEM alleles.

Cefepime is a relatively new ß-lactam antibiotic that received FDA approval in 1996. Because our in vitro evolution method accurately reproduces the natural evolution of the TEM ß-lactamases, we can use the same method to predict how the TEM ß-lactamases will evolve in the future. In this study we have used our in vitro evolution method to determine whether a new phenotype, resistance to cefepime, will arise through natural evolution of the TEM ß-lactamases.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

Esherichia coli strain DH5E [F^- 80dlacZM15 (lacZYA-argF)U169 endA1 recA1 hsdR17(r - m+) deoR thi-1 phoA supE44 ^- gyrA96 relA1 gal-] (GIBCO, Gaithersburg, MD) was used as the host for all plasmids. Plasmid pACSE3 (BARLOW and HALL 2002 ) was used as the vector for cloning and expressing TEM alleles. Site-directed mutagenesis was performed according to the manufacturer's instructions using the Quick-Change kit from Stratagene (La Jolla, CA).

In vitro mutagenesis, cloning, sequencing, and determination of minimum inhibitory concentrations (MICs) of antibiotics were as previously described (BARLOW and HALL 2002 ). Selection of evolved mutants was as previously described (BARLOW and HALL 2002 ) except that only one drug, cefepime, was used as the primary selective agent, and only ampicillin was used as a selective agent to ensure that a preexisting resistance phenotype had been retained. Briefly, the TEM-1 gene was mutagenized using the error-prone polymerase Mutazyme (Stratagene) in a PCR reaction under conditions that generated an average of two mutations per molecule. Mutagenized genes were cloned into pACSE3, a low-copy-number vector based on plasmid pACYC184 (BARLOW and HALL 2002 ) and transformed into E. coli strain DH5-E, and the resulting libraries were grown in increasing concentrations of cefepime. Each of the eight mutant libraries was passaged twice through twofold serial dilutions of cefepime (64–0.5 µg/ml), once through ampicillin (64 µg/ml), and then once again through a dilution series of cefepime (64–0.5 µg/ml). For each library, cells taken from the highest concentration of cefepime at which growth occurred were used to inoculate the next culture. Multiple passages through cefepime ensured that cefepime-resistant alleles dominated the culture. The single passage through ampicillin required that the mutant alleles also maintain the ability to confer resistance to ampicillin, a commonly used ß-lactam.

Our in vitro evolution method differs from other methods in that (a) it permits precise control over the average number of mutations introduced into each molecule; (b) it utilizes an enzyme whose mutagenic spectrum is very similar to the spontaneous mutagenic spectrum of E. coli; (c) it introduces mutations at random positions throughout each molecule; (d) it selects at clinically realistic drug concentrations; and (e) it uses more realistic selection regimes, including selection for maintenance of important wild-type resistance phenotypes, e.g., continued resistane to ampicillin. While other approaches have incorporated various elements of our method (STEMMER 1994 ; VAKULENKO et al. 1998 ; BLAZQUEZ et al. 2000 ), none have incorporated all.

Following selection, plasmid from each library was prepared from the highest concentration of cefepime at which growth occurred. Those plasmid preparations were then used as starting material for the next round of mutagenesis.

	RESULTS AND DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

Resistance to an antibiotic can be quantified by determining the MIC of the antibiotic on a bacterial strain. The MIC is the lowest concentration of the antibiotic that can completely block microbial growth. An MIC of 32 µg/ml is the breakpoint for clinical resistance to cefepime (NATIONAL COMMITTEE FOR CLINICAL LABORATORY STANDARDS 2001), but the highest reported level of TEM-conferred cefepime resistance is an MIC of 8 µg/ml (PERILLI et al. 2000 ; REBUCK et al. 2000 ). While it is possible that the TEM ß-lactamases lack the potential to evolve high levels of activity toward cefepime, it is also possible that because of the recent introduction and relatively low use of cefepime the TEM ß-lactamases have not had time to evolve resistance to this antibiotic.

To determine which amino acid substitutions, if any, increase the specificity of the TEM ß-lactamases for cefepime, we created, in E. coli K12 strain DH5-E, eight independent libraries of mutant TEM alleles and selected for increased cefepime resistance using an approach similar to the one previously described (BARLOW and HALL 2002 ). Four cycles of mutagenesis and selection were required to obtain cells that could grow in cefepime at a concentration of at least 64 µg/ml. Plasmid from those cells was harvested and retransformed into naïve DH5-E to eliminate any host selection that had occurred during growth in cefepime or ampicillin. Ten individual colonies from each transformation were screened for resistance to cefepime by disk diffusion test as previously described (BARLOW and HALL 2002 ) and a single cefepime-resistant colony from each transformation was chosen. In each set among the 10 colonies, only one resistant phenotype was present. This pattern is consistent with our earlier observation (BARLOW and HALL 2002 ) that during selection a single clone comes to dominate the population. The final result was therefore eight independent alleles, each of which was derived from an independent library of mutated TEM-1 alleles, and each of which was the result of four rounds of mutation and selection.

The MICs of several ß-lactam antibiotics for each chosen allele are shown in Table 1. All alleles increased cefepime resistance relative to TEM-1. When TEM-1 is expressed, cefepime has an MIC of 0.5 µg/ml whereas, when the evolved alleles are expressed, cefepime has an MIC of 16–256 µg/ml. Expression of allele 8 confers the highest level of resistance. Although the libraries were never exposed to ceftazadime or aztreonam, all alleles confer significantly higher resistance than TEM-1 to those antibiotics. This demonstrates that resistance to cefepime indirectly selects resistance to both aztreonam and ceftazadime.

The sequences of the evolved TEM alleles were determined and the differences between those sequences and the sequence of TEM-1 are shown in Table 2. All eight alleles have an amino acid substitution at position 164 and six of the eight alleles contain the substitution R164H. Six alleles also contain the substitution I173V. The high frequency of those amino acid substitutions suggests that both are important for resistance to cefepime. The importance of those two substitutions is further suggested by the presence of both substitutions within the most resistant allele, allele 8. However, neither of those substitutions exist in allele 1. Because allele 1 confers the second highest level of cefepime resistance, there are clearly multiple potential pathways for TEM-1 to evolve cefepime resistance.

Because natural mutations generally occur one at a time and because our in vitro mutagenesis technique simultaneously introduces multiple substitutions, it is possible to recover phenotypes from in vitro mutagenesis that would never arise in nature (HALL 2002 ). For example, if two substitutions are individually deleterious, but advantageous when they are together, they would probably not go to fixation in nature, but they might well be recovered through in vitro evolution procedures that introduce mutations at a high frequency. To verify that the mutations we recovered in the best allele, allele 8, can also be recovered from natural evolution, we determined whether a pathway exists between TEM-1 and allele 8 in which the three substitutions found in clone 8 can be introduced one at a time such that each additional substitution confers an increase in cefepime resistance.

To do that, we introduced each of the amino acid substitutions into the TEM-1 gene by site-directed mutagenesis. Of the three substitutions, we found that the substitution R164H conferred the largest increase in cefepime resistance (Table 1) with an MIC of 2 µg/ml. To the R164H allele we added the substitution I173V or R178S to produce two distinct double-mutant alleles, R164H/I173V and R164H/R178S. Of those, we found that the R164H/I173V allele conferred the greatest increase in resistance, giving a cefepime MIC of 32 µg/ml. To that allele we added the substitution R178S and found that cefepime resistance increased to a final MIC of 256 µg/ml. Because sequential addition of those three single substitutions confers an increase in cefepime resistance with each addition, there is a pathway through which natural selection can create a TEM allele that confers high levels of cefepime resistance.

Other enzymes such as the OXA and metallo ß-lactamases confer resistance to cefepime; however, neither of those families of resistance genes are as widely dispersed or as common as the TEM genes. The substitution R164H has already been observed in numerous TEM alleles that are found in clinical isolates (http://www.lahey.org/studies/webt.htm). To our knowledge, the substitutions I173V and R178S have never been observed in natural TEM alleles. Because only three amino acid substitutions are required to greatly increase the activity of the TEM enzymes toward cefepime, and because one of those mutations already exists in nature, it seems likely that the naturally occurring TEM ß-lactamases will evolve the ability to confer cefepime resistance.

While we have used in vitro evolution to make specific predictions about how cefepime resistance will arise through modification of the TEM enzymes, the in vitro evolution method we have developed can also be used to make predictions about how other resistance genes may evolve in nature. The resources available to pharmaceutical companies would enable a more thorough analysis of the evolutionary pathways through which antibiotic resistance may arise. In addition to predicting the ways in which specific enzymes can evolve the ability to confer resistance to different antibiotics, pharmaceutical companies could also use this method to test the effectiveness of using different antibiotic combinations to inhibit the evolution of resistance. For example, while it is clear that selection for resistance to cefepime indirectly selects TEM alleles that confer high levels of resistance to aztreonam and ceftazadime, it is possible that the TEM enzymes are not capable of conferring high levels of resistance to cefepime and cefotaxime at the same time. Combination of those two antibiotics may increase the time required for resistance to those antibiotics to evolve. Again, the resources available to industry would permit screening a sufficient number of libraries (at least 100) to determine whether combining cefotaxime and cefepime would effectively preclude resistance to either drug from arising. The in vitro evolution method we have developed will allow pharmaceutical companies to assess the relative ease or difficulty with which resistance to a new antibiotic or combination of antibiotic will arise. This information will help physicians and hospitals to develop intelligent strategies for prescribing antibiotics before they ever see resistance to them arise. The method we developed will also give pharmaceutical companies the ability to implement structural information about how resistance evolves into designs for future antibiotics that they will design and produce.

	ACKNOWLEDGMENTS

This study was supported by grant GM-60761 from the National Institutes of Health.

Manuscript received July 22, 2002; Accepted for publication December 20, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

BARLOW, M. and B. G. HALL, 2002  Predicting evolutionary potential: in vitro evolution accurately reproduces natural evolution of the TEM ß-lactamase. Genetics 160:823-832.[Abstract/Free Full Text]

BLAZQUEZ, J., M. I. MOROSINI, M. C. NEGRI, and F. BAQUERO, 2000  Selection of naturally occurring extended-spectrum TEM beta-lactamase variants by fluctuating beta-lactam pressure. Antimicrob. Agents Chemother. 44:2182-2184.[Abstract/Free Full Text]

CHANAL, C., R. BONNET, C. DE CHAMPS, D. SIROT, and R. LABIA et al., 2000  Prevalence of beta-lactamases among 1,072 clinical strains of Proteus mirabilis: a 2-year survey in a French hospital. Antimicrob. Agents Chemother. 44:1930-1935.[Abstract/Free Full Text]

COHEN, M. L., 2000  Changing patterns of infectious disease. Nature 406:762-767.[Medline]

COURVALIN, P. and P. TRIEU-CUOT, 2001  Minimizing potential resistance: the molecular view. Clin. Infect. Dis. 33(Suppl. 3):S138-S146.

HALL, B. G., 2002  Predicting evolution by in vitro evolution requires determining evolutionary pathways. Antimicrob. Agents Chemother. 46:3035-3038.[Abstract/Free Full Text]

KNOX, J. R., 1995  Extended-spectrum and inhibitor-resistant TEM-type beta-lactamases: mutations, specificity, and three-dimensional structure. Antimicrob. Agents Chemother. 39:2593-2601.[Medline]

LIVERMORE, D. M., 1996  Are all beta-lactams created equal? Scand. J. Infect. Dis. Suppl. 101:33-43.[Medline]

MCGOWAN, J. E., JR., 2001  Economic impact of antimicrobial resistance. Emerg. Infect. Dis. 7:286-292.[Medline]

MEDEIROS, A. A., 1997  Evolution and dissemination of beta-lactamases accelerated by generations of beta-lactam antibiotics. Clin. Infect. Dis. 24:S19-S45.

NATIONAL COMMITTEE FOR CLINICAL LABORATORY STANDARDS, 1999 Performance Standards for Antimicrobial Susceptibility Testing: Ninth Informational Supplement. NCCLS Document M100–S9, National Committee for Clinical Laboratory Standards, Wayne, PA.

Performance Standards for Antimicrobial Susceptibility Testing: Supplemental Tables.. (2001) NCCLS Document M 100–S11,(National Committee for Clinical Laboratory Standards, Wayne, PA).

PERILLI, M., B. SEGATORE, M. R. DE MASSIS, M. L. RICCIO, and C. BIANCHI et al., 2000  TEM-72, a new extended-spectrum beta-lactamase detected in Proteus mirabilis and Morganella morganii in Italy. Antimicrob. Agents Chemother. 44:2537-2539.[Abstract/Free Full Text]

REBUCK, J. A., K. M. OLSEN, P. D. FEY, K. L. BERGMAN, and M. E. RUPP, 2000  In vitro activities of parenteral beta-lactam antimicrobials against TEM-10-, TEM-26- and SHV-5-derived extended-spectrum beta-lactamases expressed in an isogenic Escherichia coli host. J. Antimicrob. Chemother. 46:461-464.[Abstract/Free Full Text]

STEMMER, W. P. C., 1994  Rapid evolution of a protein in vitro by DNA shuffling. Nature 370:389-390.[Medline]

VAKULENKO, S. B., B. GERYK, L. P. KOTRA, S. MOBASHERY, and S. A. LERNER, 1998  Selection and characterization of beta-lactam-beta-lactamase inactivator-resistant mutants following PCR mutagenesis of the TEM-1 beta-lactamase gene. Antimicrob. Agents Chemother. 42:1542-1548.[Abstract/Free Full Text]

YAN, J. J., S. M. WU, S. H. TSAI, J. J. WU, and I. J. SU, 2000  Prevalence of SHV-12 among clinical isolates of Klebsiella pneumoniae producing extended-spectrum beta-lactamases and identification of a novel AmpC enzyme (CMY-8) in Southern Taiwan. Antimicrob. Agents Chemother. 44:1438-1442.[Abstract/Free Full Text]

This article has been cited by other articles:

				W.-J. Chen, F. Delmotte, S. R. Cervera, L. Douence, C. Greif, and M.-F. Corio-Costet At Least Two Origins of Fungicide Resistance in Grapevine Downy Mildew Populations Appl. Envir. Microbiol., August 15, 2007; 73(16): 5162 - 5172. [Abstract] [Full Text] [PDF]

				E. Bantinaki, R. Kassen, C. G. Knight, Z. Robinson, A. J. Spiers, and P. B. Rainey Adaptive Divergence in Experimental Populations of Pseudomonas fluorescens. III. Mutational Origins of Wrinkly Spreader Diversity Genetics, May 1, 2007; 176(1): 441 - 453. [Abstract] [Full Text] [PDF]

				K. J. Welsh, M. Barlow, F. C. Tenover, J. W. Biddle, J. K. Rasheed, L. A. Clark, and J. E. McGowan Jr. Experimental Prediction of the Evolution of Ceftazidime Resistance in the CTX-M-2 Extended-Spectrum Beta-Lactamase Antimicrob. Agents Chemother., March 1, 2005; 49(3): 1242 - 1244. [Abstract] [Full Text] [PDF]

				B. G. Hall In Vitro Evolution Predicts that the IMP-1 Metallo-{beta}-Lactamase Does Not Have the Potential To Evolve Increased Activity against Imipenem Antimicrob. Agents Chemother., March 1, 2004; 48(3): 1032 - 1033. [Abstract] [Full Text] [PDF]

				S. J. Salipante, M. Barlow, and B. G. Hall GeneHunter, a Transposon Tool for Identification and Isolation of Cryptic Antibiotic Resistance Genes Antimicrob. Agents Chemother., December 1, 2003; 47(12): 3840 - 3845. [Abstract] [Full Text] [PDF]

				M. Barlow and B. G. Hall Experimental Prediction of the Evolution of Cefepime Resistance From the CMY-2 AmpC {beta}-Lactamase Genetics, May 1, 2003; 164(1): 23 - 29. [Abstract] [Full Text] [PDF]