DNA Microarray for Detection of Macrolide Resistance Genes

A DNA microarray was developed to detect bacterial genes conferringresistance to macrolides and related antibiotics. A databasecontaining 65 nonredundant genes selected from publicly availableDNA sequences was constructed and used to design 100 oligonucleotideprobes that could specifically detect and discriminate all 65genes. Probes were spotted on a glass slide, and the array wasreacted with DNA templates extracted from 20 reference strainsof eight different bacterial species (Streptococcus pneumoniae,Streptococcus pyogenes, Enterococcus faecalis, Enterococcusfaecium, Staphylococcus aureus, Staphylococcus haemolyticus,Escherichia coli, and Bacteroides fragilis) known to harbor29 different macrolide resistance genes. Hybridization resultsshowed that probes reacted with, and only with, the expectedDNA templates and allowed discovery of three unexpected genes,including msr(SA) in B. fragilis, an efflux gene that has notyet been described for gram-negative bacteria.

INTRODUCTION

Top

Introduction

Resistance to macrolides and related antibiotics (macrolides-lincosamides-streptogramins [MLS]) is of great concern becausethese drugs are commonly used to treat many different infectioussyndromes and because this resistance is spreading among gram-positiveand gram-negative bacteria, including strains isolated fromlife-threatening infections such as pneumonia, sepsis, endocarditis,and meningitis. Different classes of genes coding for MLS resistancehave been described, and their nucleotide sequences are availablein public databases (22). Although macrolide resistance is presentworldwide, patterns and mechanisms of resistance may vary widelyin different geographic areas, leading to different therapeuticstrategies for infective syndromes, such as community-acquiredpneumonia (15, 16, 19).

Detection of single bacterial genes (e.g., antibiotic resistancegenes or species-specific genes) in diagnostics and in epidemiologicalstudies is typically carried out by PCR, whereas DNA microarrayshave been developed to perform a large number of different hybridizationexperiments simultaneously on a single membrane or glass substrate.They are well-suited to comprehensively investigate and quantitativelycompare the expression levels of a large number of genes, butthey can also be easily used in qualitative studies to detectselected DNA sequences (7, 8, 21). To assist epidemiologicalstudies on the genetics of macrolide resistance in clinicalisolates, a method based on DNA microarrays was developed tocomprehensively assess the presence of MLS genes in bacterialgenomes.

MATERIALS AND METHODS

Top

Materials and Methods

Database construction and probe design. The sequences of MLS resistance genes were retrieved from publicdatabases and comparatively analyzed to avoid redundancy. Thefile containing the selected sequences in multi-FASTA format(http://www.compbio.ox.ac.uk/faq/format_examples.shtml) wasused to generate a database to be searched by Array Designer2.0 software (Premier Biosoft, Palo Alto, CA). Probes, 40 to60 nucleotides in size, with a melting temperature of 83 ±1°C, were designed to specifically target each gene of thedatabase. Oligonucleotide probes generated by the software werechecked for homology to unrelated sequences present in publicdatabases, and, when possible, two probes for each gene weredesigned for the array.

Construction of microarray slides. Oligonucleotide probes were synthesized by MWG Biotech (Munich,Germany), with a C₆ amino linker to allow better binding tothe slide. Epoxy-modified glass slides (Pan-Epoxy slides; MWGBiotech) and a four-head pin ring spotting apparatus (GMS 417arrayer; Genetics MicroSystems, Woburn, MA) were used. Probeswere spotted in at least three replicates at a concentrationof 30 pmol/µl in 20% dimethyl sulfoxide and 0.1% Tween20. Resulting spots had a diameter of 80 to 120 µm.

Template DNA extraction, labeling, and hybridization. Genomic DNA was extracted from a 10-ml bacterial culture harvestedin exponential phase, according to a published protocol (4).For staphylococci, 20 U of lysostaphin was added to the lysissolution. One microgram of template DNA, in a reaction volumeof 25 µl, was labeled with the fluorescent cytosine analogCy5 (Amersham Biosciences, Piscataway, NJ) by random primingusing 40 U of Klenow DNA polymerase, with a Cy5/dCTP ratio of1. Ten microliters of the labeled DNA was brought to a volumeof 14 µl in hybridization buffer (3x SSC [1x SSC is 0.15M NaCl plus 0.015 M sodium citrate], 30 mM HEPES, pH 8, 0.3%sodium dodecyl sulfate, 5x Denhardt's solution), containingtRNA of Saccharomyces cerevisiae (Sigma, St. Louis, Mo.) at1.5 mg per ml. After 2 min of denaturation at 100°C and10 min at room temperature, the 14-µl mix was layeredon the slide and hybridized for 1 h at 55°C. Slides werewashed twice for 5 min in 2x SSC-0.1% sodium dodecyl sulfateat 65°C and then twice for 5 min in 1x SSC at room temperatureand twice for 5 min in 0.2x SSC.

Data analysis. Microarray slides were read using a GMS 428 array scanner (GeneticsMicroSystems, Woburn, MA). Data were acquired using GenePixPro 5.0 software (Axon Instruments, Union City, CA) and managedwith Microsoft Excel. For each spot, median pixel intensitywas assessed, and background signal was subtracted. To controlfor congruity of results obtained with replicate spots of aprobe, the mean fluorescence intensity and the standard deviation(intraprobe standard deviation) was calculated for each probe.If, for a probe, the intraprobe standard deviation was higherthan the mean fluorescence intensity, hybridization resultswere considered negative. The standard deviation of the meanfluorescence intensity of all probes of the microarray was alsocalculated. A probe was considered positive when its fluorescenceintensity was higher than the mean fluorescence intensity ofall probes plus 1 standard deviation.

Bacterial strains. We hybridized total DNA from 20 bacterial strains carrying referenceMLS resistance genes (Table 1).

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

RESULTS AND DISCUSSION

Top

Results and Discussion

Probes for macrolide resistance genes. A database which included 65 nonredundant macrolide resistancegenes published in GenBank was selected (Tables 2 to 4). Geneswere identified by accession number, since in some cases twoor more genes with different sequences share the same name.One hundred oligonucleotide probes were designed and spottedon the microarray slide to allow differential detection of the65 selected MLS genes. Probes for ribosomal methylation genesand their positions in the coding sequence are reported in Table2, probes for efflux genes in Table 3, and probes for genescoding for esterases, nucleotidyltransferases, phosphotransferases,acetyltransferases, and hydrolases in Table 4.

View this table:
[in this window]
[in a new window]
TABLE 2. Probes for ribosomal methylation genes

View this table:
[in this window]
[in a new window]
TABLE 4. Probes for genes coding for esterases, nucleotidyltransferases, phosphotransferases, acetyltransferases, and hydrolases

View this table:
[in this window]
[in a new window]
TABLE 3. Probes for efflux genes

Microarray hybridization. Microarray slides were tested by hybridizing DNA templates extractedfrom 20 strains belonging to eight different species and knownto harbor 29 different MLS genes (Table 1). All of the probesdesigned to be specific for the 29 MLS genes reacted with thepredicted DNA templates, allowing validation of a total of 48probes (Table 5). Three unexpected results were also obtained:(i) the DNA of Bacteroides fragilis V503 reacted with probemsrSA-31, (ii) the DNA of Staphylococcus aureus BM12235 reactedwith probe vgaAv-39, and (iii) the DNA of S. aureus BM4611 reactedwith probes ermC-19 and ermC-20 (Table 5).

View this table:
[in this window]
[in a new window]
TABLE 5. Hybridization results^a

Identification of additional MLS genes in control strains. Microarray data indicating the presence of unexpected MLS genesin control strains were confirmed by DNA sequencing of the entireopen reading frame, using templates obtained by PCR, as previouslydescribed (23). In B. fragilis strain V503, carrying the methylasegene erm(FU), sequence data indicated the concomitant presenceof an efflux gene identical to msr(SA) (100% identity at theDNA level) of S. aureus (GenBank accession no. AB013298). Themsr(SA) gene is considered typical of Staphylococcus spp. andhas never been found in gram-negative bacteria. In S. aureusstrain BM12235, carrying the major facilitator streptograminefflux gene vga(B) and the streptogramin acetyltransferase genevat(B), it was possible to identify also the presence of vga(A)v,an ATP-binding transporter gene which is commonly associatedwith vga(B) and vat(B) (11, 12). DNA sequence analysis showedthat vga(A)v of BM12235 was essentially identical (99% identityat the DNA level) to vga(A)v of S. aureus BM3327 (GenBank accessionno. AF186237). In S. aureus strain BM4611, carrying the lincomycinnucleotidyltranferase gene lnu(A), an associated methylase geneof the erm(C) class was found, with up to 90% identity at thenucleotide level with several erm(C) genes present in GenBank.

Conclusions. This work provides detailed information for construction ofa simple and powerful tool to investigate the genetic basisof macrolide resistance in bacterial isolates. Careful analysisof DNA sequences deposited in public databases allowed compilationof a list of 65 bacterial genes encoding resistance to macrolidesand related drugs. Oligonucleotide DNA microarrays designedto detect these 65 genes in bacterial genomes were producedand used to test a collection of strains carrying well-characterizedMLS genes. Results provided both (i) validation of the microarraychip and (ii) proof of concept that the microarray approachis effective in detecting associations of MLS genes not necessarilyinferred by the resistance phenotype. Unlike other DNA microarraysdeveloped to detect the most common resistance genes (8, 21),this one, by its comprehensive approach, is well-suited forsurveillance studies specific for MLS resistance, where characterizationof the resistance genotype is sought. This DNA microarray couldsignificantly contribute to molecular epidemiology studies byallowing simultaneous testing for the presence of known MLSgenes and in particular could help to define and understandthe clustering of different MLS genes in genetic elements andgenomes.

ACKNOWLEDGMENTS

We thank Annalisa Pantosti for advice and all researchers listedin Table 1 for kindly providing control strains.

The work was funded in part by grants from Istituto Superioredi Sanità, from the University of Siena (PAR), and fromMIUR (FIRB, RBAU01X9TB).

FOOTNOTES

* Corresponding author. Mailing address: LAMMB, Università di Siena, Policlinico Le Scotte/V Lotto, Viale Bracci, 53100 Siena, Italy. Phone: 39 0577 233299. Fax: 39 0577 233334. E-mail: pozzi{at}unisi.it.

REFERENCES

Top