Whole-Genome Mapping: a New Paradigm in Strain-Typing Technology

Abstract

Strain-typing technology in support of outbreak identification and resolution has evolved from phenotypic analysis, such as serology and biotypes, to much-more-robust molecular genetic approaches, such as pulsed-field gel electrophoresis (PFGE) and whole-genome sequencing. Whole-genome mapping (WGM) has been recently applied to subtyping analysis, and it bridges the gap between PFGE (∼20 bands sorted by size) and whole-genome sequencing. WGM utilizes restriction site analysis but arranges 200 to 500 bands in the order they appear on the chromosome. WGM is able to quickly and cost-effectively generate high-resolution, ordered whole-genome maps of bacteria.

INTRODUCTION

Probably the most common subtyping procedure used today for differentiating bacterial strains causing community or institutional outbreaks is pulsed-field gel electrophoresis (PFGE), which was introduced in 1984 by researchers at Columbia University (1). PFGE provided a method for characterizing outbreak strains that was more robust and reproducible than serotyping, biotyping, phage typing, multilocus enzyme electrophoresis, and other commonly used methods. In PFGE, microbial DNA is extracted and subjected to one or more enzymes that cut the DNA at specific base pair sites (restriction sites), leaving the DNA in a number of strands of differing lengths. The resulting fragmented DNA is applied to a gel, and voltage is periodically alternated among three equidistant directions. As the DNA moves through the gel, large fragments must reconfigure to move slowly through the gel while smaller DNA bands move quickly. The resulting bands appear on the gel by size, with the larger bands at the top of the gel and the sequentially smaller bands toward the bottom. The result is a “DNA fingerprint” or barcode which can be used to compare the band pattern with a control and with multiple suspected outbreak strains to help determine if the isolates tested are indistinguishable by PFGE or if they are different, as defined by standard criteria (2, 3).

Multilocus sequence typing (MLST) is a highly discriminating molecular technique that has been successfully applied to strain typing but was initially used to study population genetics. MLST characterizes species by genetic sequencing of 6 to 8 specific conserved housekeeping genes (loci). MLST was developed initially for Neisseria meningitidis (4) and applied to Streptococcus pneumoniae (5) but has been used to characterize many other microbes (6–8). MLST involves PCR amplification followed by DNA sequencing of the specific allele profiles. Different genes are analyzed in each genus and species tested. Because it is a sequence-based technology, strong bioinformatics support is required.

Another discriminating molecular technique using specific loci within the prokaryotic and eukaryotic genome is multiple-locus variable-number tandem-repeat analysis (MLVA) (9–11). Many microbial genes have loci that contain tandem, repetitive DNA sequences that can distinguish strains based on the number of repeat units present. These loci can be amplified by PCR analysis and sized by capillary electrophoresis. The calculated numbers of repeats of the loci (alleles) are combined into a string which consists of integers and is referred to as the MLVA profile.

Genetic sequencing of the entire genome offers a more detailed analysis of an organism's genetic structure and produces a comprehensive DNA fingerprint that can be analyzed and interpreted only by using sophisticated bioinformatics software. Nevertheless, this method (Sanger sequencing) has been applied to subtyping analysis (12). Newer, high-throughput, and more-efficient methods of whole-genome sequencing, called next-generation sequencing, offer the possibility to further this process and provide more-affordable applications now relegated to PFGE and the more-expensive and cumbersome Sanger sequencing (13–15), although extensive bioinformatics computing power will still be required for analysis.

A relatively new genomic technology that is neither PFGE nor genome sequencing was introduced in the 1990s as “optical mapping” (16), which produces a similar barcode-like genetic map of restriction sites but has the depicted sites arranged in the order they occur in the genome rather than by size distribution in a gel (17). In short, mapping is not the same as sequencing or PFGE. This technology has matured into a highly useful tool for strain typing, comparative genomics, and whole-genome sequence assembly and is currently referred to as “whole-genome mapping” (WGM). WGM, as a strain-typing tool, provides more-accurate and -robust information than PFGE while at the same time distinguishing easily between related and unrelated strains associated with an outbreak. PFGE results in around 20 fragments resolved by molecular size and not in the order they occur on the chromosome. Whole-genome maps, with 200 to 500 restriction fragments for a typical bacterial isolate, are produced de novo, independent of sequence information and without PCR amplification, and provide a significantly higher level of discriminatory power than does PFGE.

Little data comparing WGM with MLST as a typing tool are available, although WGM does not involve genome sequencing. However, as part of an ongoing collaborative study (E. Zentz, OpGen, Inc., personal communication), MVLA, PFGE, and WGM data from isolates of two food-borne outbreaks were compared. In the first outbreak of Salmonella enterica serovar Enteritidis, MLVA and PFGE data determined that all isolates in the outbreak were clonal and WGM data confirmed these results. For the second outbreak of Escherichia coli O157:H7, MLVA data agreed with PFGE data, showing all isolates to be clonal. However, WGM was able to further differentiate the isolates, and 2 isolates in the outbreak exhibited differences in the genome that were undetected by both PFGE and MLVA. One isolate had a major duplication in the genome that was 102 kb in size. The other isolate had a 19-kb deletion in the genome relative to those of the other isolates in the outbreak. This further illustrates the sensitivity that WGM provides when comparing very closely related isolates.

HOW WGM WORKS

Bacteria, parasites, yeasts, and other fungi have been successfully analyzed by WGM (18–22). The organisms are gently lysed to obtain genomic high-molecular-weight DNA that is, on average, 250 kb or greater. The lysis procedure is a simple approach that consists of washing the cells, centrifuging, creating spheroplasts with specific enzymes, and then lysing, all without the need for vortexing or other methods that typically cause DNA shearing. The DNA is placed on a microfluidics device containing two components, (i) a charged glass surface that binds DNA and (ii) a channel-forming device with microchannels that allow the DNA strands to flow across the surface. As the DNA flows through the channels, each long molecule is stretched and immobilized on the glass surface and can be visualized as a microscopic white line, each line coming from a single piece of chromosome or DNA molecule (Fig. 1). Next, the selected restriction enzyme is applied to the surface to cut the DNA at specific restriction sites. A gap is formed at each cut site, resulting from the DNA being stretched and under some tension while adhered to the glass slide. The resulting DNA fragments are visualized and measured following fluorescent staining and are then analyzed and assembled (mapped) using an automated system (Fig. 2). This barcode represents the cut sites, and the white gaps between the bars represent the distances between the cut sites (Fig. 3). The system's software documents the restriction fragment sizes, which are converted into kilobase data equivalent to fragment size. Therefore, for single-molecule maps, the sizes of the restriction fragments are known, as are the number of fragments and their exact order in the genome. The automated system collects thousands of these molecules for analysis using a filter that selects only those molecules that are >150 kb in size as a single-molecule map set. Up to nine complete maps can be produced within 24 h on the instrument. Current technology allows completion of the entire WGM process from DNA extraction to complete bioinformatic analysis in 48 h. This is, of course, longer than the time required for PFGE analysis, but WGM provides more-robust information and is a different technology.

Fig 1.

Immobilize and digest. Single DNA molecules are flowed through microfluidic channels and immobilized on a charged glass surface. The immobilized DNA is digested, maintaining the fragment order. Genomic DNA, captured as single DNA molecules from random breakage of intact chromosomes, is loaded into microchannels, immobilized electrostatically, and then digested with a restriction endonuclease. Digestion reveals cleavage sites, with the restriction fragment order maintained for each molecule.

Fig 2.

Measure and assemble. The DNA fragments are stained with fluorescent dye; fragment length is proportional to fluorescence intensity. By overlapping fragment patterns, the single-molecule maps are assembled to produce a whole-genome map.

Fig 3.

Whole-genome restriction map. Vertical lines represent restriction sites; distances between lines represent fragment sizes.

The whole-genome map is assembled and printed as a barcode, not of nucleotides, as with genomic sequences, but as restriction assemblies of the fragment size patterns. In fact, this WGM process is robust enough to serve as an adjunct to whole-genome sequencing to detect and correct potential misassemblies in sequence data. Thus, the WGM can be used as a validation scaffold to orient and order a sequence contig (overlapping DNA segments from a specific region) placement and to correct sequence misassemblies because the restriction fragment pattern from WGM should match the restriction fragment patterns obtained from a DNA sequence.

STRAIN-TYPING APPLICATIONS OF WHOLE-GENOME MAPPING

The value of WGM as a strain-typing tool in E. coli outbreaks has been documented (23–26). In May 2011, Germany began to experience a serious food-borne outbreak due to a strain of E. coli O104:H4 (27) characterized by an increased number of cases of hemolytic-uremic syndrome. Initially, the hypothesis for the illness was that the outbreak was due to an enterohemorrhagic E. coli (EHEC) strain. It was later shown to have been caused by an enteroaggregative E. coli (EAEC) strain that had acquired the Shiga toxin genes (28, 29). During the outbreak, four samples were submitted for WGM analysis along with two strains of E. coli O104:H4 from previous unrelated outbreaks. Nearly identical de novo, WGM from all four outbreak samples confirmed the clonality (same source) of the outbreak. In addition, WGM enabled identification of the closest relative from a 2001 outbreak, chromosomal location of the Shiga toxin (stx2) and resistance (tehA, associated with tellurite resistance) genes, and also the identification of three unique, highly conserved regions common among all outbreak isolates of this E. coli strain (30) (OpGen, Inc., unpublished data). This additional information would not have been available from PFGE, which identified only the clonality of the isolates. Furthermore, the new information provided by WGM can focus classic genetic sequencing to target the study of these new regions and structural changes.

An example of the discriminatory power of WGM is shown in Fig. 4, with two S. aureus samples that were both initially pulse field typed as USA 400 shown. When these two were subjected to WGM, significant differences were noted between them. In Fig. 4, a de novo WGM of the clinical sample is compared to an in silico WGM of the reference strain (USA400 MW2), created by importing sequence data directly from NCBI. The sequence is parsed for the restriction cut sites and then converted into an in silico WGM. Any gene or feature in genomes that has been annotated to Gen Bank is also imported into the map analysis software along with the sequencing data. Therefore, with WGM, one can visualize the important genetic elements that were missed by PFGE, i.e., a staphylococcal cassette chromosome mec element (SCCmec) responsible for methicillin resistance, VS-alpha, and a region encoding the Panton-Valentine leukocidin (PVL) toxin were able to be resolved with WGM. One may conclude that this particular clinical sample was a methicillin-susceptible strain and not a methicillin-resistant strain, as predicted by PFGE.

Fig 4.

PFGE versus whole-genome mapping. MSSA, methicillin-susceptible S. aureus; MRSA, methicillin-resistant S. aureus.

On 21 October 2010, toxigenic Vibrio cholerae O1, serotype Ogawa, biotype El Tor, was identified by the National Laboratory of Public Health of the Ministry of Public Health and Population in Haiti (31). WGM of strains from this widespread outbreak compared with strains from other locations demonstrated distinct genomic diversity in different regions of the chromosome. In one CDC study (32) with PFGE, five strains of V. cholerae were found to be indistinguishable from one another using both NotI and SfiI digestion. When these same five strains were analyzed by WGM, differences between these five strains, including one strain with a 6-kb insertion that PFGE did not detect, were able to be detected.

Rapidly placing related strains together in the context of outbreak recognition and resolution must be the hallmark of any typing system. WGM can quickly discriminate between closely related strains, providing accurate and actionable information for public health epidemiologists and hospital infection control professionals. One advantage of WGM compared to PFGE is that the map data can be correlated directly to sequence data so that genetic markers for resistance or virulence can be recognized early and an appropriate intervention can be started.

Because the WGM surveys thousands of locations along the microbial chromosome, highly accurate structural detail is provided for each mapped strain. The software available with WGM provides map-based similarity clusters, or dendrograms, with more rigor than can be achieved by PFGE, which provides only a limited number of restriction bands ordered by size, and may detect important structural changes and mobile genetic motifs that can be missed even by next-generation sequencing.

ACCESSING THE TECHNOLOGY

WGM has been available for over 10 years, but its move from the research laboratory to the strain-typing laboratory is gaining momentum. Commercial WGM is provided by one source, OpGen, Inc., a DNA analysis company located in Gaithersburg, MD (www.opgen.com). OpGen receives samples through its MapIt Service laboratory for WGM analysis from worldwide sources. Alternatively, the Argus whole-genome mapping system and consumables are available to bring WGM processing and analysis to individual laboratories. The Argus system includes a MapCard processor, where the digestion and staining occur, and the Argus Mapper, which contains the fluorescence microscope and charge-coupled-device (CCD) camera used for collection of the high-resolution DNA images. The reagents required include a menu of enzymes, fluorescent stain, and MapCard, in addition to MapSolver software for organizing projects and performing data analysis. Training is provided in the use of the Argus system, and additional online support is available. The current list price of a complete system for public health applications approaches $225,000, and the cost per map ranges from $150 to $300.

Implications of whole-genome mapping for public health and health care.

A genomic approach to strain typing is clearly the gold standard for high-accuracy isolate characterization and PFGE, as utilized by the PulseNet program of the Centers for Disease Control and Prevention, and has been universally adopted to recognize and resolve outbreaks (33, 34). As sequencing technology has advanced to faster, more-economical platforms, the instrumentation and necessary bioinformatics have become more available for laboratories to utilize sequencing for chromosomal analysis with more discriminatory power. Very often, there is increased interest in applying additional discriminatory power to a group of PFGE-related isolates to further characterize and resolve an outbreak.

What will be the next-generation gold standard after PFGE? Competing resources are often the limiting factor for both public health and health care laboratories to adopt newer sequencing or mapping technology. WGM has been proven to be a highly discriminating and efficient method for strain typing of outbreak samples. The method provides more-detailed genome-wide information and can be used as an adjunct to PFGE to detect important structural changes in bacterial genomes. WGM also provides the critical bridge for public health applications to next-generation sequencing with a robust and reproducible genetic scaffold for subsequent sequencing of novel sites along the chromosome. Indeed, WGM may be a stand-alone strain-typing method following further research and validation to define its discriminatory limits. Work is under way to clarify the WGM resolution limits of “indistinguishable” and “different” strains. WGM is not yet ready to replace or compete with PFGE as a strain-typing tool, but it can be used to enhance the detail necessary to fully characterize outbreak strains. Once fully validated against the current PFGE gold standard, WGM may play a valuable role in outbreak response in public health as well as in health care-acquired infections and outbreaks.

Advancing technology usually outstrips our ability to interpret and apply its results correctly, and the tsunami of genetic data from sequence analysis will certainly bring with it the challenge of accurate interpretation and application. WGM brings a new paradigm for consideration that is worthy of review. Data analysis from traditional PFGE and data from the newest next-generation technologies can be integrated with WGM to provide a thorough, complete analysis of samples with all platforms. It is highly likely that the public health community may need more than one tool to fully appreciate the need for full outbreak analysis.

Biography

J. Michael Miller received his graduate degree from the University of Texas Health Science Center at San Antonio in 1976, after which he pursued fellowship training in clinical microbiology in the 2-year postdoctoral residency program in Medical and Public Health Laboratory Microbiology at the Centers for Disease Control and Prevention (CDC) in Atlanta, GA. After completion of his training in clinical microbiology, Dr. Miller joined the staff at the CDC and remained there for 35 years prior to his retirement in 2011. During his time at the CDC, he worked in the former Division of Training, where he developed courses and taught microbiology diagnostic procedures around the world; he has served as the Branch Chief of the Epidemiology and Laboratory Branch in the Hospital Infections Program, as the Chief of the Laboratory Response Branch in what is now the Division of Preparedness and Emerging Infections, and as the Associate Director for Laboratory Science in the National Center for Prevention, Detection, and Control of Infectious Diseases. Most recently, Dr. Miller was the Associate Director for Laboratory Science for the National Center for Emerging and Zoonotic Infectious Diseases. While at the CDC, he established the first biofilm laboratory in public health and also standardized pulsed-field gel electrophoresis (PFGE) as the CDC's primary staphylococcal typing tool. Currently, he is the owner and director of Microbiology Technical Services, LLC, a private clinical microbiology consulting service for laboratories throughout the United States. Dr. Miller developed and currently manages ClinMicroNet, a worldwide listserv for doctoral-level microbiology laboratory directors. He established and developed the CDC's position in administering BioWatch for the Department of Homeland Security at that program's inception and restructured and oversaw the CDC's Laboratory Response Network for 5 years. Dr. Miller has authored two textbooks in clinical microbiology, numerous book chapters, and more than 100 peer-reviewed scientific articles. Dr. Miller has been active in numerous capacities in both the American Society for Microbiology and the Clinical and Laboratory Standards Institute and was one of the founding members of the National Laboratory Training Network. He is the former dean of ASM's American College of Microbiology and for 6 years served on the Board of Governors of the American Academy for Microbiology. He is the recipient of numerous awards in clinical microbiology. These include the ASM Clinical Microbiology Leadership Award, the ASM Public Health Microbiology Award, the ASM Founders Distinguished Service Award, and the 2012 Pasteur Award from the Illinois Society for Microbiology.