This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowReprints and Permissions
Right arrow Copyright Information
Right arrow Books from ASM Press
Right arrow MicrobeWorld
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Lerner, C. G.
Right arrow Articles by Beutel, B. A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Lerner, C. G.
Right arrow Articles by Beutel, B. A.

 Previous Article  |  Next Article 

Antimicrobial Agents and Chemotherapy, July 2005, p. 2767-2777, Vol. 49, No. 7
0066-4804/05/$08.00+0     doi:10.1128/AAC.49.7.2767-2777.2005
Copyright © 2005, American Society for Microbiology. All Rights Reserved.

Novel Approach to Mapping of Resistance Mutations in Whole Genomes by Using Restriction Enzyme Modulation of Transformation Efficiency

Claude G. Lerner,* Stephan J. Kakavas, Christian Wagner, Richard T. Chang, Philip J. Merta, Xiaoan Ruan, Randy E. Metzger, and Bruce A. Beutel

Global Pharmaceutical Research Division, Abbott Laboratories, Abbott Park, Illinois 60064

Received 24 June 2004/ Returned for modification 19 September 2004/ Accepted 24 March 2005


arrow
ABSTRACT
 
Restriction enzyme modulation of transformation efficiencies (REMOTE) is a method that makes use of genome restriction maps and experimentally observed differences in transformation efficiencies of genomic DNA restriction digests to discover the location of mutations in genomes. The frequency with which digested genomic DNA from a resistant strain transforms a susceptible strain to resistance is primarily determined by the size of the fragment containing the resistance mutation and the distance of the mutation to the end of the fragment. The positions of restriction enzyme cleavage sites immediately flanking the resistance mutation define these parameters. The mapping procedure involves a process of elimination in which digests that transform with high frequency indicate that the restriction enzyme cleavage sites are relatively far away from the mutation, while digests that transform with low frequency indicate that the sites are close to the mutation. The transformation data are compared computationally to the genome restriction map to identify the regions that best fit the data. Transformations with PCR amplicons encompassing candidate regions identify the resistance locus and enable identification of the mutation. REMOTE was developed using Haemophilus influenzae strains with mutations in gyrA, gyrB, and rpsE that confer resistance to ciprofloxacin, novobiocin, and spectinomycin, respectively. We applied REMOTE to identify mutations that confer resistance to two novel antibacterial compounds. The resistance mutations were found in genes that can decrease the intracellular concentration of compounds: acrB, which encodes a subunit of the AcrAB-TolC efflux pump; and fadL, which encodes a long-chain fatty acid transporter.


arrow
INTRODUCTION
 
Resistance to antibacterial agents is a serious medical problem, particularly in hospitals, long-term care facilities, day care centers, and intensive care units (10, 11, 22, 39, 51). A critical step in addressing this growing dilemma is the ability to rapidly determine the genetic mechanisms behind highly resistant clinical isolates. Some cases can be predictably traced to specific mutations at previously identified genetic loci, while others may involve completely unexpected mutations in other parts of the bacterial genome. Understanding the genetic basis for resistance is essential for formulating new strategies to combat highly resistant clinical isolates and also advances our understanding of this dynamic process and its relationship to emerging resistance in various clinical settings (1, 31, 51). Additionally, some antibacterial agents have been found to target different proteins in different organisms. For example, in gram-negative bacteria first-step resistance to many quinolones is found to occur in DNA gyrase; in contrast, first-step resistance in gram-positive strains instead involves topoisomerase IV (14, 21). Discovering such differences in antibacterial targets and mechanisms of action can facilitate development of agents with an improved antibacterial spectrum.

Resistance mutations can also provide valuable information about the target of novel antibacterial compounds with unknown mechanisms of antibacterial activity (3, 13, 52). However, resistance mutations can arise in genes other than those encoding the protein target of the antibacterial compound (29, 30, 41); therefore, several resistant isolates may need to be analyzed to identify the actual target protein. Discovering the target of a novel antibacterial compound with an unknown mechanism of action can greatly facilitate preclinical development. Knowing the target can enable rapid improvement in therapeutic and pharmacological profiles through medicinal chemistry optimization assisted by structure-based methods such as X-ray crystallography and protein nuclear magnetic resonance.

Significant advances have been made in the ability to identify and detect mutations (53), including enzymatic and chemical methods (5, 9, 50), as well as genotyping via microarrays (20, 44, 54). While these methods hold promise for future application on a genomewide scale, they currently have not advanced to the stage of practical use for rapid and robust identification of resistance mutations in bacterial genomes. Classical methods of genetic mapping in bacteria, while generally robust, are time-consuming, provide low-resolution information, and often require a collection of well-characterized reference strains and/or other specialized reagents. To increase the throughput and shorten the time needed to identify locations of resistance mutations, we developed a simple method based on the heterogeneity of restriction endonuclease cleavage fragments in digests of chromosomal DNA. The method was developed using Haemophilus influenzae strains with known resistance mutations in gyrA, gyrB, and rpsE that confer resistance to ciprofloxacin (Cipr), novobiocin (Novr), and spectinomycin (Sptr), respectively. We have termed the method "REMOTE," for "restriction enzyme modulation of transformation efficiencies." We also describe application of REMOTE for discovery of mutations that confer resistance to novel antibacterial compounds with unknown mechanisms of action.


arrow
MATERIALS AND METHODS
 
Genomic DNA digestion and H. influenzae transformation. The cetyltrimethylammonium bromide method (4) was used for isolation of genomic DNA from H. influenzae with the addition of 0.0l ml of RNase A (50 µg/ml) and incubation at 37°C for 15 min, prior to the second phenol extraction. DNA was isolated from H. influenzae NP200 (40) derivatives resistant to the novel antibacterial compound A-344583 (NP200-583.3R) or A-84568 (NP200-568.1R) or H. influenzae Super8 (kindly provided by Jane K. Setlow, Brookhaven National Laboratory), a strain that is multiply antibiotic resistant, including mutations in the gyrA, gyrB, and rpsE genes (TIGR and NCBI RefSeq accession numbers HI1264, NP_439419; HI0567, NP_438724; and HI0795, NP_438954) that confer resistance to ciprofloxacin, novobiocin, and spectinomycin, respectively (Table 1).


View this table:
[in this window]
[in a new window]
  		TABLE 1. Bacterial strains used in this study

A panel of 34 restriction enzymes (listed in Fig. 4) was selected based on average fragment sizes determined by in silico genome digests. Enzymes that cut both frequently (in the 200- to 500-bp size range) and infrequently (in the 3,000- to 5,000-bp range) were included in the panel. If the mutation is in a region where a frequent cutter doesn't cut or an infrequent cutter does cut, then the number of possible candidate locations is greatly reduced. For H. influenzae, the majority of enzymes in the panel yielded fragments that averaged between 500 and 2,000 bp. Multiple enzymes compensate for differences in enzyme cleavage efficiencies and sequence variations, including cases where the mutation creates or eliminates a restriction site.



View larger version (31K):
[in this window]
[in a new window]
  		FIG. 4. Transformation frequencies obtained with H. influenzae NP200 host cells using restriction enzyme digests of H. influenzae strain Super8, which contains a Cipr mutation in gyrA, a Novr mutation in gyrB, and a Sptr mutation in rpsE. Restriction enzyme names are indicated as labels above the bars, whose heights correspond to the observed transformation frequencies. The ordinate indicates the background-corrected number of transformants obtained with the digests. Enzyme classifications determined by method A or B are indicated at the top of the graphs (F, full-effect enzymes; M, moderate-effect enzymes; N, no-effect enzymes). Asterisks indicate which classification method correctly identified the mutation as the top coordinate.

Samples of the purified DNA (1 to 2 µg) were completely digested with a 10-fold excess of restriction enzyme according to the manufacturer's directions. The digests were then purified by phenol extraction, and the concentration of digested DNA was determined fluorometrically. Two hundred nanograms of purified digested DNA was mixed with susceptible H. influenzae NP200 cells made competent for transformation as previously described (6). The transformation mixture (1 ml) was then incubated at 37°C for 30 min. Five milliliters of supplemented brain heart infusion broth (sBHI) was then added, and the cells were incubated at 37°C for 1 h. Aliquots of 0.001, 0.01, and 0.1 ml were plated onto sBHI agar plates containing 0.03 µg/ml ciprofloxacin (Cip), 5 µg/ml novobiocin (Nov), 50 µg/ml spectinomycin (Spt), 30 µg/ml A-344583, or 50 µg/ml A-84568. The plates were incubated overnight at 37°C to select for growth of resistant colonies.

REMOTE analysis. The transformation frequencies for the digests are determined and entered into a computer program along with the genome restriction map derived from the genome sequence (i.e., the H. influenzae Rd genome sequence, RefSeq accession number NC_000907) (16). The program generates a list of genome coordinates and corresponding scores that measure the "fit" of the transformation data to the restriction map surrounding each coordinate. The scores are then sorted to identify the coordinates that that best fit the transformation data. PCR (42) amplification products (~1.5 kbp) of the top candidate regions are generated with genomic DNA from the resistant strain and then are tested by transformation to see which one confers resistance in a susceptible strain. The exact location and nature of the mutation are then determined by sequencing the amplification product of the positive locus from the resistant mutant and susceptible parent.

The transformation data are used to classify the enzymes into three categories according to the extent to which they decrease transformation efficiency (no effect, moderate effect, and full effect). The decision on how to bin enzymes into these three categories was made empirically using the three control mutations. For these three mutations, two methods of binning, A and B, give the most accurate identification of the loci. The percentage of the maximal transformation frequency is first calculated for each enzyme relative to digests that yield the highest transformation efficiencies. The data are then assessed by either method A or method B. In method A, full-effect, moderate-effect, and no-effect enzymes are defined as those with percent maximal transformation frequencies of <0.1%, 0.1% ≤ x ≤ 1.5%, and >1.5%, respectively. In method B, full-effect, moderate-effect, and no-effect enzymes are defined as those with percent maximal transformation frequencies of <1.0%, 1.0% ≤ x ≤ 2.5%, and >2.5%, respectively.

In addition to fragment length and the distance of a mutation from the end of a fragment, some species of bacteria, such as H. influenzae, contain specific DNA uptake signal sequences (USS) that are needed for efficient transformation (45, 46). In these organisms, restriction digests that result in the nearest USS sequences being cut off of the mutation-bearing fragment give rise to low transformation frequencies (similar to a restriction site being close to the mutation). The analysis is done slightly differently in these organisms. In this case, the genome is scanned to identify regions that (i) contain sites that would produce fragments that lack USS sequences and/or clusters of sites for enzymes that decrease transformation frequencies but (ii) do not contain clusters of sites for enzymes that do not reduce transformation frequencies. This process yields a refined list of candidate regions in the genome, one of which most likely contains the mutation.

For a coordinate of the genome sequence, the program calculates both the length of predicted restriction fragments for restriction sites flanking the coordinate and the distance of the coordinate to the sites. By comparing the calculated length and distances to user-defined parameters, the program classifies enzymes as being predicted to have a moderate effect, full effect, or no effect on lowering transformation efficiencies. The values of the user-defined parameters are determined experimentally from control experiments. The program then assesses whether the observed transformation effect matches the calculated effect for each input enzyme, and the numbers of correct and incorrect matches at the coordinate are tallied and stored. The program then moves down the sequence 10 bp and repeats the analysis. The analysis is done every 10th bp rather than at each base to reduce computation time. After completing the entire genome, a list is generated that is sorted to identify the locations that contain the most correct enzyme effect matches. Scanning the list for successive coordinates that differ by more than 500 bp identifies the genome coordinates that best match the experimental data. The output can be limited by visualizing only those coordinates for which the calculated and empirical transformation data match by a specified percentage. A cutoff of 80% yields files of manageable size (<20 MB).

The user-defined parameters for analysis of the sequence-derived restriction map include two windows that surround the test location. One is the size of a small window (sw) within which mutations would be so close to the restriction site that the transformation frequency would nearly be zero. Enzymes with sites within this window are thus designated as full-effect enzymes. A larger window (lw) is also defined beyond which the mutation would be so far away from the restriction site that transformation is predicted to occur with high frequency. Enzymes that cleave between the small and large windows are expected to give rise to low but detectable numbers of transformants and thus are defined as moderate-effect enzymes.

Two other user-defined parameters take into account the dependence of transformation efficiency on total fragment length (L; full/moderate-effect enzyme fragment size cutoff [fmfrag] and no-effect enzyme fragment size cutoff [nfrag]). Enzymes that generate fragments encompassing the test coordinate that are too short to yield moderate or high transformation frequencies (L < fmfrag) are placed in the full-effect category. Enzymes that generate fragments so long that they are not expected to decrease the transformation frequency (L > nfrag) are classified as no-effect enzymes. Enzymes that generate fragments of intermediate length (fmfrag < L < nfrag) are expected to noticeably, but not completely, decrease transformation. Such enzymes are classified as having a moderate effect on transformation. The hierarchy used to assign an enzyme to a category is full effect > moderate effect > no effect. For example, if a long fragment (L > nfrag, no effect) has an end that is close to the mutation (<sw, full effect) the enzyme is placed into full-effect category. The algorithm can also take into account the presence or absence of USS DNA uptake sequences on the fragment.

For our studies with H. influenzae, we used both enzyme effect classification methods (A and B above) with parameter settings of sw at 100 bp, lw at 400 bp, fmfrag at 1,200, and nfrag at 2,500 bp. The program was run in two modes, one of which took into account USS sequences. In the analyses, all five resistance loci were found as the top candidate region in one of the four analyses (method A or B, with and without consideration of USS sequences). In general, we recommend running the computer program four times, including methods A and B, with and without USS sequences being taken into account. While for most mutations, one of these four methods will identify the proper locus as the top candidate region, we would conservatively recommend assessing each of the top three to five regions for each of the four methods. This can be done by testing PCR products from each region for the ability to confer the mutant phenotype by transformation.

Resistant mutant isolation. Preferably resistant mutants should be isolated as colonies on agar plates; however, we were developing a method for large-scale isolation of resistant mutants and found selections in liquid cultures to be operationally more expedient. Suspensions of H. influenzae NP200 cells were prepared in sBHI from freshly streaked agar plates. Five-hundred-milliliter cultures were inoculated to give a starting optical density at 600 nm (OD600) of 0.002 and incubated at 37°C with shaking to an OD600 of 0.2, which corresponds to a density of 1 x 108 to 5 x 108 CFU/ml. Drug was added to a final concentration of 1x, 2x, or 4x the MIC, and incubation was resumed for 4 to 8 h. The cultures were then diluted 1:10 into fresh medium with drug and incubated until the OD600 returned to 0.2. This process of cycling was continued, as the drug-sensitive cells were gradually replaced by a resistant mutant subpopulation. After each cycle, aliquots of the cultures were plated to screen for contamination and cryopreserved at –80°C in 20% glycerol.

The resistant cultures were plated out onto sBHI plates in parallel with the parent strain. Twenty colonies were purified by consecutively plating two times on drug plates followed by three rounds of plating on drug-free media. Subsequently, stable resistant isolates were confirmed by plating on selective media. The isolates were then tested alongside the parent strain for drug resistance levels in liquid culture by broth microdilution (32). Isolates exhibiting resistance profiles most specific for the drug were tested by ribotyping to confirm isogenicity with the NP200 parent strain. Ribotyping was performed with the Riboprinter microbial characterization system (DuPont Qualicon, Wilmington, DE) according to the manufacturer's instructions.

H. influenzae FLUSKO strain construction. To get a better understanding of the relationship between transformation frequency and restriction fragment size or mutation distance from a fragment end, an H. influenzae strain was created in which a USS (45, 46) was adjacent to the resistance mutation. Since premature termination of the fadL gene was found to confer resistance to A-344583 (see Results and Discussion; Table 2), the fadL gene was disrupted by incorporation of a stop codon adjacent to an uptake-specific sequence (USS) within the fadL coding region. This strain was termed "FLUSKO" (Table 1), for "FadL uptake sequence knockout." To generate the strain, a 700-bp region of the fadL gene from H. influenzae NP200 was amplified by PCR using the upstream primer 5'-TAAAACGGCACAGTTTTCCAAAGTGCGGTAACGGGTGGCGTTTATGTTG-3' and downstream primer 5'-TACCATCTTCGAAGCTAGCG-3'. The USS is shown in bold letters, and the stop codon is underlined. The PCR fragment was purified by spin-column gel filtration and quantitated fluorometrically. Chemically competent H. influenzae NP200 cells were transformed using 200 ng of the PCR fragment, and aliquots were plated onto sBHI agar containing 30 µg/ml of compound A-344583 and incubated overnight in a CO2 incubator at 37°C. Genomic DNA from four colonies was prepared, and the fadL gene was PCR amplified using N-terminal and C-terminal primers. The sequence of the engineered region was confirmed using an upstream sequencing primer. Replacement of the wild-type fadL sequence with the PCR fragment incorporates the USS sequence after nucleotide 205 of fadL and puts the stop codon in frame.


View this table:
[in this window]
[in a new window]
  		TABLE 2. Novel antibacterial compounds A-344583 and A-84568 for which resistance mutations in H. influenzae were identified by REMOTE


arrow
RESULTS AND DISCUSSION
 
In several organisms transformation frequencies have been found to be proportional to the length of DNA fragments carrying a mutation (7, 25-28, 57). We therefore hypothesized that the physical location of restriction enzyme sites relative to the location of a mutation could be predicted from differences in transformation frequencies observed with restriction enzyme digests of genomic DNA. We postulated that enzymes that cleave close to a mutation would yield digests that transform with lower frequencies than those that cleave at sites farther from the mutation. In essence, the transformation data enables construction of a crude restriction map of the region surrounding the mutation. This crude map is compared with the actual restriction map derived from the genome sequence to identify the physical locus that best fits the experimental data.

Examination of the genome restriction map for regions that (i) contain clusters of cleavage sites for enzymes that decrease transformation frequencies, but (ii) do not contain cleavage sites for enzymes that do not reduce transformation frequencies, provides a short list of candidate regions in the genome, one of which most likely contains the mutation (Fig. 1). A computer program is used to compare the empirical data to the restriction map of the entire bacterial genome in order to identify the locations that best fit the data (see Materials and Methods for program details).



View larger version (26K):
[in this window]
[in a new window]
  		FIG. 1. Diagram of the basic principle of REMOTE. Genomic DNA from a mutant strain is digested individually with restriction enzymes (e.g., A, B, C, and D). The frequency with which the digests transform a nonmutant strain to the mutant phenotype is assessed. Digests that transform with low frequency indicate the restriction sites are in close proximity to the mutation (*). Conversely, digests that transform with high frequency indicate the sites are far from the mutation (denoted by {zac007055077star}). The mutation is indicated by an inverse white circle ({zac007055077gph1}). For a hypothetical genomic segment containing 10 genes, the data fit best to gene 3, which contains sites for both low-transforming digests (C and D) in close proximity and the sites for the high-transforming digests (A and B), which are further away. The other nine genes do not have clusters of the sites for both low-transforming digests that are also far from sites for the high-transforming digests.

The dependence of transformation frequency on fragment length and mutation distance from fragment ends in any transformable species can be established empirically with control strains containing mutations in known locations, using PCR products of various lengths containing a mutation in the middle of the fragment, and by using PCR products of constant length with the mutation at different positions from the end of the fragment. To assess the dependence of transformation frequency on the proximity of a mutation to the end of a fragment in H. influenzae, PCR was used to generate a 1-kbp fragment that contained a resistance mutation at various distances from the end of the fragment. PCR products containing a ciprofloxacin resistance missense mutation in the gyrA gene at various distances from the end of the fragments were transformed into a sensitive strain (H. influenzae NP200, Table 1). As the distance of the mutation from the end of the fragment decreased, the transformation efficiency decreased, with a significant decrease in transformation frequencies occurring when the mutation was less than 50 bp from the end of the fragment (Fig. 2).



View larger version (23K):
[in this window]
[in a new window]
  		FIG. 2. Relationship between transformation frequency and the distance of a resistance mutation from the end of a DNA restriction fragment in H. influenzae. Ciprofloxacin-sensitive NP200 host cells were transformed to ciprofloxacin resistance with 1-kb PCR products containing a ciprofloxacin resistance gyrA mutation located at various distances from the end of the fragments. The PCR products were amplified from the ciprofloxacin-resistant H. influenzae gyrA strain Super8. To roughly approximate the conditions used in transformations with genomic digests, the transformation contained 1 ng of PCR product and 200 ng of digested NP200 genomic DNA.

To assess the dependence of the transformation frequency on fragment length, we used a strain (H. influenzae FLUSKO; Table 1) engineered to contain a USS immediately adjacent to a mutation (A-344583 resistance in fadL) so each restriction fragment containing the resistance mutation would also contain a USS. Thus, all restriction fragments would be able to gain entry into the cell via the USS sequence, so decreases in transformation would not be due to the lack of USS-mediated DNA uptake. Since the mutation is located at various locations along the length of the genomic restriction fragments, a subset that contained the mutation more than 200 bp away from an end were used for the analysis to minimize potentially confounding end effects. Figure 3 shows the dependence of transformation efficiency on DNA fragment length observed with restriction enzyme digests of DNA from the FLUSKO strain. As the fragment length approaches ~2,500 bp, the transformation efficiency decreases 10-fold. Below ~1,500 bp, the transformation efficiency decreases 100-fold.



View larger version (33K):
[in this window]
[in a new window]
  		FIG. 3. Dependence of transformation frequency on the length of restriction fragments carrying a resistance mutation. H. influenzae NP200 host cells were transformed to A-344583 resistance with 200 ng of restriction enzyme digests of H. influenzae strain FLUSKO genomic DNA. To minimize the impact of end effects, data are shown only for enzymes that contain the mutation at least 200 bp from the ends of the restriction fragments.

Identification of known resistance mutations in H. influenzae by REMOTE. To develop the REMOTE mutation identification method in H. influenzae, we used a multiply resistant strain (Super8; Table 1) that is resistant to eight antibiotics, including ciprofloxacin, novobiocin, and spectinomycin. The ciprofloxacin, novobiocin, and spectinomycin transformation frequency profiles for a set of restriction enzyme digests of Super8 genomic DNA transformed into the susceptible host strain NP200 are shown in Fig. 4. The enzyme effect categories determined by methods A and B are also indicated on the figure. The asterisk indicates which method identified the correct mutation locus as the top hit. Method A successfully identified the gyrA locus for the Cipr transformation data and the rpsE locus for the Sptr data. Method B also successfully identified gyrA as the Cipr locus, and gyrB as the Novr locus. The patterns of transformation frequencies vary between the three resistance phenotypes. Most of the digests give rise to Sptr or Novr transformants, but less than half of the digests yield Cipr transformants. In particular, the HaeIII, HhaI, SspI, and RsaI digests yield over 10,000 Novr or Sptr transformants, but no Cipr transformants, while the Hpy99I digest yields over 10,000 Cipr transformants but fewer than 100 Sptr transformants. As discussed above, these experimentally observed transformation frequencies enable deduction of a low-resolution map of the relative location of restriction sites surrounding the mutations. For example, since they have little effect on Novr and Sptr transformation frequencies, sites for HaeIII, HhaI, SspI, and RsaI are predicted to be far from these resistance mutations. In contrast, digests with these same enzymes did not yield Cipr transformants, so the Cipr mutation is predicted to be close to these sites. Similarly, sites for Hpy99I are predicted to be relatively far from the Cipr mutation since many Cipr transformants were observed with this digest. Conversely, the Sptr mutation is predicted to be located in a region that is close to Hpy99I sites, since few Sptr transformants were obtained with this digest.

An illustration of how transformation frequencies obtained with a battery of restriction enzyme digests correlate with the physical restriction map around the corresponding resistance mutation is shown in Fig. 5. In this example, the Cipr, Novr, and Sptr transformation data are compared with three loci, gyrA, gyrB, and rpsE, which are known to contain the corresponding resistance mutations. As expected, the restriction map of the gyrA locus fits the Cipr transformation data better than the maps of the gyrB or rpsE loci. Digests that do not significantly decrease the Cipr transformation frequency (no-effect enzymes, N) correspond to longer restriction fragments, all of which contain a USS and arise by cleavage at sites relatively far from the mutation. Enzyme digests that moderately decrease Cipr transformation frequency (moderate-effect enzymes, M) are closer to the mutation and shorter in length. Enzymes that result in the most profound decrease in transformation frequency (full-effect enzymes, F) cluster close to the site of the mutation, and most do not contain USS uptake sequences. The gyrA locus clearly shows a correlation between the physical proximity of restriction sites to the mutation and the degree to which the transformation frequency is affected by the restriction enzyme digest. The gyrB or rpsE loci on the other hand do not exhibit such a correlation and thus do not fit the experimental Cipr data as well as the gyrA locus. For example, at the gyrB and rpsE loci, sites for the Cipr no-effect enzymes are found close to the mutation, unlike at the gyrA locus. For all three sets of transformation data, sites for full-effect enzymes are clustered close to the mutation at correct loci and sites for no-effect enzymes are found at more distal locations. The same trends can be seen for the Novr transformation data (compare the gyrB locus to the gyrA or rpsE loci) and the Sptr transformation data (compare the rpsE locus to the gyrA and gyrB loci). Note that clustering of sites for full-effect enzymes is often observed at incorrect loci. This is expected because these are sites for enzymes that frequently cut the genome and are found near most loci. Thus, they yield digests that profoundly decrease transformation frequencies of most genes.



View larger version (33K):
[in this window]
[in a new window]
  		FIG. 5. Locations of restriction enzyme cleavage sites surrounding the Cipr mutation in gyrA, the Novr mutation in gyrB, and the Sptr mutation in rpsE. For each mutation, groups of maps are shown for enzymes from the transformation effect categories, no-effect (N), moderate-effect (M), and full-effect enzymes (F). Classification method A (for details, see Materials and Methods) was used to assign the enzymes into the categories for the Cipr and Sptr transformation data, and method B was used for the Novr data, as depicted in Fig. 4. Vertical bars represent locations of restriction endonuclease cleavage sites. Horizontal lines represent the genome sequence, and vertical dotted lines represent locations of USS DNA uptake sequences. The distances (bp) from the mutations are indicated at the bottom of the panels. Only the two restriction sites that are nearest to the mutation (one on each side) are shown. Note that clustering of full-effect enzymes will not appear to be perfect, since only one of the two sites shown needs to be near enough to the mutation to significantly the transformation efficiency.

This type of analysis is not limited to a few loci, as just described; it can be extended to the entire H. influenzae genome by use of a computer program that rapidly compares the experimental and sequence-derived restriction maps. Such a computerized analysis successfully identified the correct loci gyrA, gyrB, and rpsE as the top candidate coordinates from the Cipr, Novr, and Sptr transformation data, respectively.

As was described above, the experimental transformation frequency data are used to create an approximate restriction map in which the locations of enzyme cleavage sites relative to the position of a resistance mutation are deduced from transformation frequencies obtained with genomic DNA digests. Perfect matches between the restriction map created from the experimental data and the map deduced from the genome sequence are not expected and are not necessary. For example, 4 of the 34 enzymes tested in the Novr and Sptr mapping experiment did not exhibit a correlation between the transformation-derived and sequence-derived restriction maps. However, in both cases the correct locus was identified as the top coordinate. Even with greater than 10% disagreement between the maps, REMOTE successfully identified the correct loci. In fact, the final discrimination of the correct locus is often determined by a few enzymes that do not affect the transformation frequency (no-effect enzymes).

Disagreement between the two maps can arise for several reasons, including sequence differences between the experimental strain and the published genome sequence. Such differences can be real or simply mistakes in the genome sequence data. Additionally, the presence of unknown sequences that affect DNA uptake or transformation frequencies and/or restriction enzymes that do not cut well at particular sites or have unexpected activity can also lead to inconsistencies between the maps. However, such differences are only relevant if they occur in relatively close proximity to the mutation. Differences beyond the set of restriction sites that immediately flank either side of the mutation do not affect the transformation results and thus do not adversely affect the analysis. Problems arising from differences more proximal to the mutation can be solved by using more enzymes to generate a more complete set of transformation frequency data with which to derive an experimental restriction map of the resistance locus. Optimizing the parameter values used in the computer analysis can also limit the negative effect of such differences.

Identification of mutations in H. influenzae that confer resistance to antibacterial compounds with unknown mechanisms of action. We applied the REMOTE mutation identification method to characterize mutations that give rise to resistance to novel antibacterial compounds; these compounds were identified in a high-throughput screen of the Abbott chemical repository for compounds that inhibit growth of H. influenzae. Resistant mutants were isolated for two such compounds, A-344583 and A-84568 (Tables 1 and 2). Both compounds had MICs of 16 µg/ml against the susceptible H. influenzae strain NP200. Resistant mutants with MICs of >128 µg/ml were isolated for both compounds. The mutants were not resistant to other classes of antibacterial agents, including rifampin, chloramphenicol, tetracycline, kanamycin, ciprofloxacin, and novobiocin, thus suggesting the possibility that the resistance mutations were in genes encoding the molecular targets of the compounds. However, REMOTE analysis determined the location of the resistance mutations to be in genes encoding proteins that probably decreased the intracellular concentrations of the compounds. The A-84568 resistance mutation was found to be located in the acrB gene that encodes the B subunit of the AcrAB-TolC efflux pump (56). The A-344583 resistance mutation was found in the fadL gene encoding a transport protein that facilitates uptake of long-chain fatty acids (8).

The acrB resistance mutation is a missense mutation that results in a change of an alanine at residue 569 to glutamate. This mutation apparently alters the substrate specificity of the AcrAB efflux pump to include or improve A-84568 as a pump substrate, thereby enabling survival of the bacterial cells in the presence of high concentrations of the compound. Interestingly, the mutation does not alter the MICs of other known AcrAB substrates tested, including ethidium bromide (data not shown).

In contrast, and by analogy with its known biological role in fatty acid uptake, the fadL mutation likely prevents cellular penetration of A-344583. The fadL mutation is a loss-of-function null mutation due to a nucleotide deletion that causes a reading frame shift, resulting in premature termination of translation.

Other methods for mapping mutations on a genomewide scale in bacteria have recently been reported (2, 7, 18, 48, 49). Several methods involve use of plasmid or bacteriophage libraries, which can be difficult and time-consuming to construct and can be biased, with certain sequences represented infrequently or not at all. Another method to identify resistance mutations employs production of a collection of mutagenized PCR products that cover the entire genome (7). An inherent limitation of such a method is the intrinsic mutation bias of error-prone PCR mutagenesis that leads to underrepresentation of certain types of mutations. This is probably why the method was not able to identify mutations that confer resistance to ciprofloxacin (7).

Other methods take advantage of the physicochemical differences between matched and mismatched DNA heteroduplexes to identify mutations (18, 48, 49). As some authors of such methods acknowledge, these technologies are labor intensive and could miss some mutations, including those that alter or are in close proximity to restriction sites (48, 49). The proposed solution is to repeat the experiment using a different restriction enzyme, thus further increasing the time needed for an already lengthy method. Methods that rely on proteins to selectively interact with mismatched heteroduplexes suffer from the intrinsic problem of insufficient substrate specificity and bias. Mismatch binding proteins have been found to exhibit various activities, depending on the type of nucleotide mismatch (33, 47). Another problem with these nonphenotypic methods is that only one or a few of the multiple nucleotide differences that are found are related to the mutant phenotype. As a result, additional experiments must be performed to identify the phenotypically relevant mutation. The strength of these methods lies in their ability to identify resistance mechanisms requiring multiple mutations at divergent loci, although such mutations are likely to be difficult to deconvolute phenotypically. In contrast, REMOTE requires less than 2 weeks to determine the exact nature and precise location of a resistance mutation in a bacterial genome.

Although not yet examined experimentally, a theoretical limitation of REMOTE will probably be an inability to identify large deletions or insertions. Thus, REMOTE may not identify resistance loci in some clinical isolates, since they frequently contain acquired resistance elements (31, 51).

In addition to H. influenzae, we have performed REMOTE in Bacillus subtilis, another naturally transformable species. Differences in rifampin resistance (RifR) transformation frequencies were observed with a set of genomic DNA digests prepared from a strain containing a RifR mutation in the rpoB gene (Fig. 6). Briefly, parameters were defined that enabled identification of the rpoB gene by computer analysis of transformation data, along with the established genome restriction map. We expect REMOTE to be applicable to other transformable bacterial species for which the entire genome sequence is available; currently this includes Agrobacterium tumefaciens, Borrelia burgdorferi, Brucella melitensis, Caulobacter crescentus, Clostridium perfringens, Corynebacterium glutamicum, Deinococcus radiodurans, Enterococcus faecalis, Escherichia coli, Helicobacter pylori, Lactococcus lactis, Listeria monocytogenes, Pasteurella multocida, Rickettsia prowazekii, Streptomyces coelicolor, Xanthomonas campestris pv. campestris, Yersinia pestis, Acinetobacter spp., Campylobacter spp., Mycoplasma spp., Neisseria spp., Nostoc spp., Pseudomonas spp., Salmonella spp., Streptococcus spp., and Synechocystis spp.



View larger version (21K):
[in this window]
[in a new window]
  		FIG. 6. Transformation frequencies obtained with B. subtilis BD170 host cells using restriction enzyme digests of a rifampin-resistant derivative, BD170-R5, containing a Rifr mutation in the rpoB gene. Restriction enzyme names are indicated as labels above the bars, whose heights correspond to the observed transformation frequencies.

Theoretically, the availability of the entire genome sequence is not absolutely necessary for the method, since the analysis can be performed on the available sequence data. However, the likelihood of identifying the location of the mutation is much lower. REMOTE should also be applicable to virtually any mutation that confers a selectable phenotype. It may prove useful in characterizing nonselectable mutant phenotypes, particularly in organisms that can be transformed with high efficiency.

REMOTE is a conceptually simple method that can be reduced to practice in different species and strains by empirically determining the variables that influence the transformation efficiency of a selectable phenotype conferred by mutations anywhere in the genome. Of the many possible applications for this technology, perhaps one of the most important is that it facilitates an ideal approach to the discovery of novel antibacterial agents. From a practical standpoint, researchers have had to choose between two extremes: cellular screening and molecular target screening, although pathway screens with cell extracts have also been pursued (12, 24, 35). Screening directly for inhibitors of bacterial proliferation results in compounds with antibacterial activity, but without a readily discovered molecular target to facilitate structure and/or mechanism-directed medicinal chemistry for lead optimization. Historically, this approach worked decades ago when highly evolved natural products (already partially optimized by nature) were discovered, resulting in the major classes of antibiotics. The alternative approach of screening for inhibitors of particular molecular targets essential for bacterial proliferation has been considerably less fruitful, as it is frequently difficult to optimize the enzyme inhibitory activity while also achieving penetration into cells and specificity of binding required for antibacterial activity (19, 34, 36-38, 43).

It would be desirable to combine the best of both approaches so that large numbers of novel antibacterial compounds from cellular screens can be used to rapidly identify their respective molecular targets. Methods have recently been developed to use gene expression profiling to implicate biochemical pathways affected by novel agents (15, 23). Although such methods do not directly identify the target of the antibacterial agent, they provide valuable information about the mode of activity and significantly narrow the list of potential targets. Target identification methods have also been described which make use of classical genetic mapping (13), cloning (2, 17, 52), error-prone PCR (7), antisense technology (55), and genome sequence analysis (3). Generally, these methods are not readily amenable to large-scale implementation for various reasons.

In comparison, REMOTE seems well suited to this approach to antibacterial discovery. If novel antibacterial agents from cellular screens can be used to generate resistant mutants, REMOTE should be able to rapidly identify putative molecular targets to facilitate medicinal chemical optimization of these agents. As shown here for the FadL and AcrB mutations, in some instances REMOTE will fail to identify targets, but it may still provide useful information about the route of bacterial entry or exit for the novel compound. However, in other cases, as was demonstrated by the control agents described here, mutations will be identified in the molecular target for the novel compound, thus enabling biochemical assays, X-ray crystallography, and other methods for optimization of the novel target-compound pair, starting with a compound that has antibacterial activity.


arrow
ACKNOWLEDGMENTS
 
We thank Jane K. Setlow for kindly providing the Super8 strain, Stella Z. Doktor for ribotyping the resistant isolates, and Kristin S. Lerner for editing the manuscript.

The authors also wish to thank the anonymous reviewers whose constructive criticisms helped us significantly improve the manuscript.


arrow
FOOTNOTES
 
* Corresponding author. Mailing address: Abbott Laboratories, R4CC, AP10-1, 100 Abbott Park Road, Abbott Park, IL 60064. Phone: (847) 937-7875. Fax: (847) 938-1674. E-mail: claude.lerner{at}abbott.com. Back


arrow
REFERENCES
 
    1
  1. Abbanat, D., M. Macielag, and K. Bush. 2003. Novel antibacterial agents for the treatment of serious Gram-positive infections. Expert Opin. Investig. Drugs 12:379-399.[CrossRef][Medline]
    2. 2
  2. Adrian, P. V., W. Zhao, T. A. Black, K. J. Shaw, R. S. Hare, and K. P. Klugman. 2000. Mutations in ribosomal protein L16 conferring reduced susceptibility to evernimicin (SCH27899): implications for mechanism of action. Antimicrob. Agents Chemother. 44:732-738.[Abstract/Free Full Text]
    3. 3
  3. Andries, K., P. Verhasselt, J. Guillemont, H. W. Gohlmann, J. M. Neefs, H. Winkler, J. Van Gestel, P. Timmerman, M. Zhu, E. Lee, P. Williams, D. de Chaffoy, E. Huitric, S. Hoffner, E. Cambau, C. Truffot-Pernot, N. Lounis, and V. Jarlier. 2005. A diarylquinoline drug active on the ATP synthase of Mycobacterium tuberculosis. Science 307:223-227.[Abstract/Free Full Text]
    4. 4
  4. Ausubel, F. M., R. Brent, R. E. Kingston, D. D. Moore, J. G. Seidman, J. A. Smith, and K. Struhl (ed.). 1991. Current protocols in molecular biology, p. 2.4.1-2.4.2. John Wiley & Sons, New York, N.Y.
    5. 5
  5. Babon, J. J., M. McKenzie, and R. G. Cotton. 2003. The use of resolvases T4 endonuclease VII and T7 endonuclease I in mutation detection. Mol. Biotechnol. 23:73-81.[CrossRef][Medline]
    6. 6
  6. Barcak, G. J., M. S. Chandler, R. J. Redfield, and J. F. Tomb. 1991. Genetic systems in Haemophilus influenzae. Methods Enzymol. 204:321-342.[Medline]
    7. 7
  7. Belanger, A. E., A. Lai, M. A. Brackman, and D. J. LeBlanc. 2002. PCR-based ordered genomic libraries: a new approach to drug target identification for Streptococcus pneumoniae. Antimicrob. Agents Chemother. 46:2507-2512.[Abstract/Free Full Text]
    8. 8
  8. Black, P. N., and C. C. DiRusso. 2003. Transmembrane movement of exogenous long-chain fatty acids: proteins, enzymes, and vectorial esterification. Microbiol. Mol. Biol. Rev. 67:454-472.[Abstract/Free Full Text]
    9. 9
  9. Bui, C. T., A. Lambrinakos, J. J. Babon, and R. G. Cotton. 2003. Chemical cleavage reactions of DNA on solid support: application in mutation detection. BMC Chem. Biol. 3:1.[CrossRef][Medline]
    10. 10
  10. Canton, R., T. M. Coque, and F. Baquero. 2003. Multi-resistant Gram-negative bacilli: from epidemics to endemics. Curr. Opin. Infect. Dis. 16:315-325.[Medline]
    11. 11
  11. Clark, N. M., J. Patterson, and J. P. Lynch III. 2003. Antimicrobial resistance among gram-negative organisms in the intensive care unit. Curr. Opin. Crit. Care 9:413-423.[CrossRef][Medline]
    12. 12
  12. Dandliker, P. J., S. D. Pratt, A. M. Nilius, C. Black-Schaefer, X. Ruan, D. L. Towne, R. F. Clark, E. E. Englund, R. Wagner, M. Weitzberg, L. E. Chovan, R. K. Hickman, M. M. Daly, S. Kakavas, P. Zhong, Z. Cao, C. A. David, X. Xuei, C. G. Lerner, N. B. Soni, M. Bui, L. L. Shen, Y. Cai, P. J. Merta, A. Y. Saiki, and B. A. Beutel. 2003. Novel antibacterial class. Antimicrob. Agents Chemother. 47:3831-3839.[Abstract/Free Full Text]
    13. 13
  13. Delgado, M. A., M. R. Rintoul, R. N. Farias, and R. A. Salomón. 2001. Escherichia coli RNA polymerase is the target of the cyclopeptide antibiotic microcin J25. J. Bacteriol. 183:4543-4550.[Abstract/Free Full Text]
    14. 14
  14. Drlica, K., and X. Zhao. 1997. DNA gyrase, topoisomerase IV, and the 4-quinolones. Microbiol. Mol. Biol. Rev. 61:377-392.[Abstract]
    15. 15
  15. Fischer, H. P., N. A. Brunner, B. Wieland, J. Paquette, L. Macko, K. Ziegelbauer, and C. Freiberg. 2004. Identification of antibiotic stress-inducible promoters: a systematic approach to novel pathway-specific reporter assays for antibacterial drug discovery. Genome Res. 14:90-98.[Abstract/Free Full Text]
    16. 16
  16. Fleischmann, R. D., M. D. Adams, O. White, R. A. Clayton, E. F. Kirkness, A. R. Kerlavage, C. J. Bult, J.-F. Tomb, B. A. Dougherty, J. M. Merrick, K. McKenney, G. G. Sutton, W. FitzHugh, C. A. Fields, J. D. Gocayne, J. D. Scott, R. Shirley, L. I. Liu, A. Glodek, J. M. Kelley, J. F. Weidman, C. A. Phillips, T. Spriggs, E. Hedblom, M. D. Cotton, T. Utterback, M. C. Hanna, D. T. Nguyen, D. M. Saudek, R. C. Brandon, L. D. Fine, J. L. Fritchman, J. L. Fuhrmann, N. S. Geoghagen, C. L. Gnehm, L. A. McDonald, K. V. Small, C. M. Fraser, H. O. Smith, and J. C. Venter. 1995. Whole-genome random sequencing and assembly of Haemophilus influenzae Rd. Science 269:496-512.[Abstract/Free Full Text]
    17. 17
  17. Gentry, D. R., K. A. Ingraham, M. J. Stanhope, S. Rittenhouse, J. L. Jarvest, P. J. O'Hanlon, J. R. Brown, and D. J. Holmes. 2003. Variable sensitivity to bacterial methionyl-tRNA synthetase inhibitors reveals subpopulations of Streptococcus pneumoniae with two distinct methionyl-tRNA synthetase genes. Antimicrob. Agents Chemother. 47:1784-1789.[Abstract/Free Full Text]
    18. 18
  18. Gotoh, K., M. Hata, M. Miyajima, and H. Yokota. 2000. Genome-wide detection of unknown subtle mutations in bacteria by combination of MutS and RDA. Biochem. Biophys. Res. Commun. 268:535-540.[CrossRef][Medline]
    19. 19
  19. Gu, Y. G., A. S. Florjancic, R. F. Clark, T. Zhang, C. S. Cooper, D. D. Anderson, C. G. Lerner, J. O. McCall, Y. Cai, C. Black-Schaefer, G. F. Stamper, P. J. Hajduk, and B. A. Beutel. 2004. Structure-activity relationships of novel potent MurF inhibitors. Bioorg. Med. Chem. Lett. 14:267-270.[CrossRef][Medline]
    20. 20
  20. Heller, M. J. 2002. DNA microarray technology: devices, systems, and applications. Annu. Rev. Biomed. Eng. 4:129-153.[CrossRef][Medline]
    21. 21
  21. Higgins, P. G., A. C. Fluit, and F. J. Schmitz. 2003. Fluoroquinolones: structure and target sites. Curr. Drug Targets 4:181-190.[CrossRef][Medline]
    22. 22
  22. Hosein, I. K., D. W. Hill, J. L. Jenkins, and J. T. Magee. 2002. Clinical significance of the emergence of bacterial resistance in the hospital environment. J. Appl. Microbiol. 92(Suppl.):90S-97S.
    23. 23
  23. Hutter, B., C. Schaab, S. Albrecht, M. Borgmann, N. A. Brunner, C. Freiberg, K. Ziegelbauer, C. O. Rock, I. Ivanov, and H. Loferer. 2004. Prediction of mechanisms of action of antibacterial compounds by gene expression profiling. Antimicrob. Agents Chemother. 48:2838-2844.[Abstract/Free Full Text]
    24. 24
  24. Kariv, I., H. Cao, P. D. Marvil, E. V. Bobkova, Y. E. Bukhtiyarov, Y. P. Yan, U. Patel, L. Coudurier, T. D. Chung, and K. R. Oldenburg. 2001. Identification of inhibitors of bacterial transcription/translation machinery utilizing a miniaturized 1536-well format screen. J. Biomol. Screen. 6:233-243.[Abstract/Free Full Text]
    25. 25
  25. Lataste, H., J. P. Claverys, and A. M. Sicard. 1981. Relation between the transforming activity of a marker and its proximity to the end of the DNA particle. Mol. Gen. Genet. 183:199-201.[CrossRef][Medline]
    26. 26
  26. Lau, P. C., C. K. Sung, J. H. Lee, D. A. Morrison, and D. G. Cvitkovitch. 2002. PCR ligation mutagenesis in transformable streptococci: application and efficiency. J. Microbiol. Methods 49:193-205.[CrossRef][Medline]
    27. 27
  27. Lee, M. S., C. Seok, and D. A. Morrison. 1998. Insertion-duplication mutagenesis in Streptococcus pneumoniae: targeting fragment length is a critical parameter in use as a random insertion tool. Appl. Environ. Microbiol. 64:4796-4802.[Abstract/Free Full Text]
    28. 28
  28. Lee, M. S., B. A. Dougherty, A. C. Madeo, and D. A. Morrison. 1999. Construction and analysis of a library for random insertional mutagenesis in Streptococcus pneumoniae: use for recovery of mutants defective in genetic transformation and for identification of essential genes. Appl. Environ. Microbiol. 65:1883-1890.[Abstract/Free Full Text]
    29. 29
  29. Levy, S. B. 2002. Active efflux, a common mechanism for biocide and antibiotic resistance. J. Appl. Microbiol. 92(Suppl.):65S-71S.
    30. 30
  30. Li, X. Z., and H. Nikaido. 2004. Efflux-mediated drug resistance in bacteria. Drugs 64:159-204.[CrossRef][Medline]
    31. 31
  31. Livermore, D. M. 2003. Bacterial resistance: origins, epidemiology, and impact. Clin. Infect. Dis. 36(Suppl. 1):S11-S23.[CrossRef][Medline]
    32. 32
  32. National Committee for Clinical Laboratory Standards. 1997. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically. Document no. M7-A4. National Committee for Clinical Laboratory Standards, Wayne, Pa.
    33. 33
  33. Oleykowski, C. A., C. R. Bronson Mullins, A. K. Godwin, and A. T. Yeung. 1998. Mutation detection using a novel plant endonuclease. Nucleic Acids Res. 26:4597-4602.[Abstract/Free Full Text]
    34. 34
  34. Overbye, K. M., and J. F. Barrett. 2005. Antibiotics: where did we go wrong? Drug Discov. Today 10:45-52.[CrossRef][Medline]
    35. 35
  35. Pratt, S. D., C. A. David, C. Black-Schaefer, P. J. Dandliker, X. Xuei, U. Warrior, D. J. Burns, P. Zhong, Z. Cao, A. Y. Saiki, C. G. Lerner, L. E. Chovan, N. B. Soni, A. M. Nilius, F. L. Wagenaar, P. J. Merta, L. M. Traphagen, and B. A. Beutel. 2004. A strategy for discovery of novel broad-spectrum antibacterials using a high-throughput Streptococcus pneumoniae transcription/translation screen. J. Biomol. Screen. 9:3-11.[Abstract/Free Full Text]
    36. 36
  36. Projan, S. J. 2002. New (and not so new) antibacterial targets—from where and when will the novel drugs come? Curr. Opin. Pharmacol. 2:513-522.[CrossRef][Medline]
    37. 37
  37. Projan, S. J., and P. J. Youngman. 2002. Antimicrobials: new solutions badly needed. Curr. Opin. Microbiol. 5:463-465.[CrossRef][Medline]
    38. 38
  38. Projan, S. J. 2003. Why is big Pharma getting out of antibacterial drug discovery? Curr. Opin. Microbiol. 6:427-430.[CrossRef][Medline]
    39. 39
  39. Raz, R. 2003. The clinical impact of multiresistant gram-positive microorganisms in long-term care facilities. J. Am. Med. Dir. Assoc. 4(Suppl. 3):S100-S104.[CrossRef][Medline]
    40. 40
  40. Reich, K. A., L. Chovan, and P. Hessler. 1999. Genome scanning in Haemophilus influenzae for identification of essential genes. J. Bacteriol. 181:4961-4968.[Abstract/Free Full Text]
    41. 41
  41. Ruiz, J. 2003. Mechanisms of resistance to quinolones: target alterations, decreased accumulation and DNA gyrase protection. J. Antimicrob. Chemother. 51:1109-1117.[Abstract/Free Full Text]
    42. 42
  42. Saiki, R. K., D. H. Gelfand, S. Stoffel, S. J. Scharf, R. Higuchi, G. T. Horn, K. B. Mullis, and H. A. Erlich. 1988. Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase. Science 239:487-491.[Abstract/Free Full Text]
    43. 43
  43. Sanders, W. J., V. L. Nienaber, C. G. Lerner, J. O. McCall, S. M. Merrick, S. J. Swanson, J. E. Harlan, V. S. Stoll, G. F. Stamper, S. F. Betz, K. R. Condroski, R. P. Meadows, J. M. Severin, K. A. Walter, P. Magdalinos, C. G. Jakob, R. Wagner, and B. A. Beutel. 2004. Discovery of potent inhibitors of dihydroneopterin aldolase using CrystaLEAD high-throughput X-ray crystallographic screening and structure-directed lead optimization. J. Med. Chem. 47:1709-1718.[CrossRef][Medline]
    44. 44
  44. Shi, M. M. 2002. Technologies for individual genotyping: detection of genetic polymorphisms in drug targets and disease genes. Am. J. Pharmacogenomics 2:197-205.[CrossRef][Medline]
    45. 45
  45. Smith, H. O., J. F. Tomb, B. A. Dougherty, R. D. Fleischmann, and J. C. Venter. 1995. Frequency and distribution of DNA uptake signal sequences in the Haemophilus influenzae Rd genome. Science 269:538-540.[Abstract/Free Full Text]
    46. 46
  46. Smith, H. O., M. L. Gwinn, and S. L. Salzberg. 1999. DNA uptake signal sequences in naturally transformable bacteria. Res. Microbiol. 150:603-616.[Medline]
    47. 47
  47. Smith, J., and P. Modrich. 1996. Mutation detection with MutH, MutL, and MutS mismatch repair proteins. Proc. Natl. Acad. Sci. USA 93:4374-4379.[Abstract/Free Full Text]
    48. 48
  48. Sokurenko, E. V., V. Tchesnokova, A. T. Yeung, C. A. Oleykowski, E. Trintchina, K. T. Hughes, R. A. Rashid, J. M. Brint, S. L. Moseley, and S. Lory. 2001. Detection of simple mutations and polymorphisms in large genomic regions. Nucleic Acids Res. 29:E111.
    49. 49
  49. Sokurenko, E. V. 2001. Discovering the sweeping power of point mutations using a GIRAFF. Trends Microbiol. 9:522-525.[CrossRef][Medline]
    50. 50
  50. Taylor, C. F., and G. R. Taylor. 2004. Current and emerging techniques for diagnostic mutation detection: an overview of methods for mutation detection. Methods Mol. Med. 92:9-44.[Medline]
    51. 51
  51. Tenover, F. C. 2001. Development and spread of bacterial resistance to antimicrobial agents: an overview. Clin. Infect. Dis. 33(Suppl. 3):S108-S115.
    52. 52
  52. Tsay, J. T., C. O. Rock, and S. Jackowski. 1992. Overproduction of ß-ketoacyl-acyl carrier protein synthase I imparts thiolactomycin resistance to Escherichia coli K-12. J. Bacteriol. 174:508-513.[Abstract/Free Full Text]
    53. 53
  53. Weaver, T. A. 2000. Trends in genetics: new technologies for life sciences: a trends guide, p. 36-42. Elsevier, New York, N.Y.
    54. 54
  54. Wong, C. W., T. J. Albert, V. B. Vega, J. E. Norton, D. J. Cutler, T. A. Richmond, L.W. Stanton, E. T. Liu, and L. D. Miller. 2004. Tracking the evolution of the SARS coronavirus using high-throughput, high-density resequencing arrays. Genome Res. 14:398-405.[Abstract/Free Full Text]
    55. 55
  55. Yin, D., B. Fox, M. L. Lonetto, M. R. Etherton, D. J. Payne, D. J. Holmes, M. Rosenberg, and Y. Ji. 2004. Identification of antimicrobial targets using a comprehensive genomic approach. Pharmacogenomics 5:101-113.[CrossRef][Medline]
    56. 56
  56. Yu, E. W., J. R. Aires, and H. Nikaido. 2003. AcrB multidrug efflux pump of Escherichia coli: composite substrate-binding cavity of exceptional flexibility generates its extremely wide substrate specificity. J. Bacteriol. 185:5657-5664.[Free Full Text]
    57. 57
  57. Zawadzki, P., and F. M. Cohan. 1995. The size and continuity of DNA segments integrated in Bacillus transformation. Genetics 141:1231-1243.[Abstract]

Antimicrobial Agents and Chemotherapy, July 2005, p. 2767-2777, Vol. 49, No. 7
0066-4804/05/$08.00+0     doi:10.1128/AAC.49.7.2767-2777.2005
Copyright © 2005, American Society for Microbiology. All Rights Reserved.




This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowReprints and Permissions
Right arrow Copyright Information
Right arrow Books from ASM Press
Right arrow MicrobeWorld
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Lerner, C. G.
Right arrow Articles by Beutel, B. A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Lerner, C. G.
Right arrow Articles by Beutel, B. A.

Copyright © 2009 by the American Society for Microbiology.   For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»