Patent Application
20090203032
The present invention relates to methods for using a specific method of solution phase competitive DNA hybridization, referred to as “Genome Fragment Enrichment” to identify microbial DNA sequences for determination of different sources of fecal contamination. The invention also relates to using this method for comparing bacterial genomes, and developing specific PCR primer sets to differentiate among bacterial species, strains, and sources of pollution.
Poultry farming for meat production has significantly increased in the last few decades. For example, the per capita consumption of chicken in the United states was estimated to be over 74 pounds in 2004, which represents a 200% increase in less than 20 years, according to the U.S. Department of Agriculture. As a result of this increase in production, fecal matter has become a significant byproduct of the poultry industry. Fecal matter is often used as fertilizer in the form of raw or composted manure. A potential risk arising from the disposal of poultry waste is the spread of enteric pathogens, such as Escherichia coli O157:H7, Salmonella spp., and Campylobacter spp. These pathogens can reach watersheds after rainfall, and thereby increase risks associated with recreational use of waterways. Furthermore, environmental concerns also include high nutrient loads, such as nitrate and phosphate, from runoff to streams, ponds, and ground water. Methods that can specifically detect poultry fecal pollution are therefore needed to assist in the development and evaluation of adequate management practices targeting pollution control.
Current regulatory methods used to assess microbial water quality rely on measuring the levels of culturable fecal indicator bacteria such as Enterococci and other fecal coliforms. However, the plate culture approach cannot discriminate among different among specific bacterial strains or animal sources of fecal contamination.
A limited number of studies have reported on the use of genotypic methods to identify the presence of poultry fecal contamination in surface waters. Ribotyping and rep-PCR DNA fingerprint techniques targeting E. coli isolates have been applied to discriminate among different animal fecal sources, including chicken and human fecal sources. However, the successful application of these genotypic methods depends on the development of large fingerprint databases of indicator bacterial isolates, primarily E. coli. Moreover, the use of E. coli for fecal source identification has been recently criticized in light of the abundance of secondary habitat populations that are capable of adapting to conditions outside of the animal gut and, as a result, contribute to the levels of fecal indicator bacteria in water.
Recently, Field and coworkers used library-independent methods based on ribosomal 16S rRNA gene (i.e., 16S sDNA) sequences of Bacteroides-like bacteria to discriminate between human and ruminant feces. These Bacteroides markers have been used to identify non-point sources of fecal pollution in coastal in inland waters. Analyses of bacterial rDNA sequences from chicken fecal DNA extracts suggests that chicken cecum and ileum are inhabited by a diverse bacterial community. Although the chicken fecal communities are different from cattle and human fecal microbial communities, thus far no studies have demonstrated the value of 16S rDNA sequences to design host-specific genetic markers. Moreover, to date, there are no non-16S rDNA library-independent assays that can determine the presence of chicken fecal pollution in watersheds.
Functional genes involved in host-microbial interactions may represent a good pool of targets for host-specific assays. Some of these functional genes are hypothesized to be microbial surface proteins, while others may be associated with cellular processes and metabolism. However, a limited number of studies have used genes involved in host-microbial interactions as potential fecal community markers. This is probably due to the small number of microbial genes known to be involved in host-microbial interactions and the limited sequence information for these genes.
There is a demand for accurate microbial source tracking (MST), because of language in the U.S. Clean Water Act regarding total maximum daily loads (TMDLs) and protection of supplies of drinking water. Current PCR-based MST approaches focus on various specific known DNA sequences, mostly targeting 16S rRNA (rDNA) genes, once thought to be source specific. However, validation studies are constantly uncovering exceptions and limitations with existing MST technologies. A significant part of the problem with existing 16S rDNA-based MST methods stemmed from the inability to target microorganism DNA sequences encoding for proteins directly involved in host-microbe interactions, which are expected to contain high levels of genetic variation related to survival within different animal hosts.
Many specific approaches have previously attempted to determine sources of fecal contamination in the environment. One of the most widely used techniques is a PCR-based method that identifies ruminant fecal pollution by targeting bacterial 16S rDNA sequences from Bacteroides (Bernard and Field, AEM 66:4571-4574, 2000). The present inventors have conducted ongoing validation studies of this method, and have discovered that previously described proposed ruminant specific markers can amplify rDNA from non-ruminant fecal samples collected from geographic regions outside the original watersheds sampled. By definition, these previously described PCR target regions identify cow, deer, elk, goat, sheep, and other ruminants and pseudo-ruminants. This approach is therefore less useful in watersheds impacted by more than one ruminant animal source.
While advances in DNA sequencing and computational biology allow scientists to compare entire microbial genomes and discern microorganism-specific genetic information, sequencing of multiple closely related bacterial genomes so far remains prohibitively expensive and impractical for all but a very small number of laboratories. The entire genome content of more than 238 bacterial species have so far been defined through whole genome sequencing of representative type strains, and the number of genome sequences continues to increase. While significant differences in the genome content of different species are well-established, comparisons between genomes of closely related bacteria are equally important. These comparisons can provide species and strain-specific genetic information, define metabolic pathways and virulence factors, and provide insights into capacities for host-interactions, cell-to-cell signaling, stress response, and other essential microbial cellular functions.
Current DNA-based technologies potentially capable of identifying source, species, and strain-specific genetic markers include Suppressive Subtractive Hybridization (SSH) (Diatchenko et al., PNAS 93:6025-6030, 1996). This technique uses intentionally biased PCR amplification of nucleic acid pools to enrich for unique segments of restricted DNA relative to non-target DNA. SSH has been successfully applied in several pair-wise comparative genome studies (e.g., Nguyen et al., 2004, AEM 71 2564-2575), but only on one “metagenomic” or total microbial community DNA study (Galbraith et al., 2004; Environmental Microbiology: 928-937). SSH is a negative selection process that relies on unequal PCR amplification to amplify all dissimilar sequences from two nucleic acid pools. This is achieved by adding different self-complementary flanking regions to each of two fragment pools, and inhibition of amplification of only those duplexes that re-anneal relative to new heteroduplexes that form following denaturation and reassociation of the mixture.
One of the limitations of currently available microbial source tracking (MST) methods arises from the inability of previously described techniques to target microorganism DNA sequences potentially encoding proteins directly involved in host-microbe interactions. These regions, unlike rDNA operons, are expected to retain high levels of genetic variation in microbes found in association with different animal hosts.
It is an object of the present invention to use the Genome Fragment Enrichment (GFE) method to identify species, strain and host-specific microbial DNA sequences.
It is a further object of the present invention to provide methods for identifying whether microbial DNA from a specific animal source is present in fecal-contaminated material.
It is still another object of the present invention to identify the described DNA sequences from Bacteroidales-like microorganisms.
It is another object of the present invention to develop PCR primer deoxyoligonucleotide pairs to differentiate among microorganisms and host animals with respect to origins of pollution.
Genome Fragment Enrichment (GFE) method is used to select for genomic regions that differ among different fecal metagenomes.
A competitive hybridization method is used to enrich for DNA metagenomic fragments specific to chicken fecal DNA. Most metagenomic clones were predicted to be similar to bacterial normally present in the chicken fecal microbial community. The cloned fragments were used to develop chicken-specific assays that were challenged using chicken and non-chicken DNA fecal extracts. The assays suggest that some genetic markers are conserved in bacteria present in the avian gastrointestinal tract.
Competitive DNA hybridizations were performed between chicken fecal DNA and pig fecal DNA (CP) and between chicken fecal DNA and an avian DNA composite consisting of turkey, goose and seagull fecal DNA extracts (CB) to search for chicken-specific DNA fragments.
The present invention provides a positive DNA selection approach designated Genome Fragment Enrichment (GFE) technique, and its efficient use in identifying both unique and divergent sequences in closely related microbial genomes. Two Enterococci species were initially studied, Enterococcus faecalis (ATCC# 19433) and Enterococcus faecium, (ATCC# 19434). This technique can be used for many other species of microorganisms and types of environmental samples.
Enterococci are natural inhabitants of many animal gastrointestinal tracts, and are commonly found in sewage and animal waste. Enterococci are therefore frequently used as indicators of fecal pollution in environmental waters, and for human exposure risk assessments. These bacteria are also opportunistic pathogens, and cause nosocomial infections. The complete annotated genome of E. faecalis V. 583 and a draft genome assembly of E. faecium (Joint Genome Institute) are now available, allowing for an accurate post-assay assessment of the reported initial application of the Genome Fragment Enrichment (GFE) method.
Competitive DNA hybridization is used to select for chicken-specific DNA fragments. A total of 471 non-redundant chicken metagenomic sequences were retrieved and analyzed. All of these sequences were similar to prokaryotic genes, of which more than 60% could not be assigned to previously characterized functional roles. In general times, sequences assigned characterized functional roles were associated with cellular processes (11.7%), metabolism (11.0%) and information storage and processing (13.4%). Approximately 53% of the non-redundant sequences are similar to genes present in intestinal bacterial belonging to Clostridia (20.9%), Bacteroidetes (15.0%) and Bacilli (17.3%).
Twenty five sequences from the CP and CB clone libraries were selected to develop chicken fecal-specific PCR assays. These assays were challenged against fecal DNA extracted from 21 different animal species, including mammals and birds. The results from the host-specificity studies showed that twelve of the assays had a high degree of specificity to chicken feces. In addition, three assays were specific to chicken and turkey while another four assays tested positive to more than two avian species, suggesting a broader distribution of some of the enriched gene fragments among different avian fecal microbial communities. Fecal pollution signals were detected using chicken-specific assays in contaminated water samples, although the PCR assays showed different detection limits.
The competitive DNA hybridization approach to detecting chicken-specific fecal microbial sequences can rapidly select for numerous chicken fecal metagenomic regions that can be used as potential genetic markers for fecal source tracking.
Accurate identification of fecal pollution from particular animal species and individual sources is critical to assess associated health risks and to develop management plans to protect recreational water and preserve the integrity of drinking water sources (i.e., rivers and aquifers). In the United States, animal source identification methods are being applied in the development of Total Maximum Daily Loads (TMDL) as part of the Clean Water Act requirements, and in the evaluation of best management practices.
The present inventors have discovered that it is possible to prepare a set of species-specific DNA sequences utilizing GFE with total DNA extracted from fecal samples that provide the sequence information required to develop species-specific PCR primers for identifying the origin of animal fecal pollution in natural waters. The utility of these sequences was clearly demonstrated in a reduction to practice exercise in which three sequences were randomly chosen and used to design cow-specific PCR primers for detecting the presence or absence detection methods. These sequences, and the other sequences in the set for cows, are potential targets for developing PCR primers for presence or absence detection methods, real-time quantification of fecal sources, and microarray applications for risk assessment and risk management. This technique has also been applied to identify fecal contamination from chicken and human species, and to differentiate fecal pollution from these sources relative to cattle, horse, sheep, goat, pig, whitetail deer, Canadian goose, seagull, turkey, and other animals that potentially contribute to fecal pollution in a natural water source.
The present invention accelerates the identification of DNA sequences from one microorganism relative to another. For example, Enterococcus faecalis-specific DNA sequences were identified by using GFE to compare E. faecalis and E. faecium genomic DNA, and enrich for E. faecalis genome-specific DNA fragments. The two microorganisms compared, however, can be of any species, strain, or isolate if necessary.
Experiments conducted with Enterococci yielded 300 probable genome-specific sequences. Genome specificity was confirmed for 225 of these DNA sequences with a comparative sequence analysis using BLAST and BLAT algorithms. E. faecalis genome-specific sequences ranged from genes encoding phage related proteins to putative surface-exposed proteins, and even detected short regions of variation embedded in highly conserved rrn sequences. Thus, this method confirms the use of comparative genomics to recognize DNA loci that can be used as indicators of fecal pollution and to identify microorganism-specific genetic markers.
The present invention makes it possible, using molecular methods, to discriminate among clinically relevant species, to study the ecology of environmentally relevant microorganism species, and to identify microorganism-specific genetic markers for stress responses, virulence, carbon utilization, and cell-to-cell communication pathways.
Isolation of previously uncharacterized sequences from a microbial fecal community was made possible with the development of a DNA sorting method called Genome Fragment Enrichment (GFE). This technique is widely applicable to developing species and strain-specific PCR primers and probes, as well as to discovering novel virulence factors, use in computational toxicology, characterization of microbial communities, development of new exposure indicators, and development of methods for environmental monitoring of microbial water quality.
Genome Fragment Enrichment (GFE) uses competitive solution hybridization to obtain DNA fragments that are present in one pool of fragments but not another (as shown in
To identify DNA targets for microbial source tracking (MST), a metagenomic approach was used (that is, compared DNA pools were from total fecal microbial community DNA). The technical challenge was to determine a way to simultaneously compare thousands of genomes isolated from fecal samples, and identify discriminatory DNA sequences from microorganisms that have not previously been cultured or characterized.
In an initial metagenomic application of the invention, cow-specific sequences were obtained by comparing the metagenomic DNA extracts derived from cow and pig fecal samples using genome fragment enrichment (GFE). GFE uses solution phase competitive nucleic acid hybridization to achieve enrichment for target molecules, as does the second step in the previously described RNA-based method for analysis of microbial gene expression Selective Capture of Transcribed Sequences (SCOTS) (Graham et al., 1999, PNAS96:11554-11559). SCOTS allows for the selective capture of bacterial cDNA molecules from total cDNA prepared from infected cells or tissues in a first step, using hybridization to biotinylated, bacterial, genomic DNA. These are previously well-described nucleic acid manipulation methods that are applied differently in each analysis method. Major key changes were required to use competitive nucleic acid hybridization for the DNA analysis method invented, Genome Fragment Enrichment.
Fundamental differences between SCOTS and GFE are that GFE identifies regions of DNA variation, rather than differentially expressed genes (as RNAs). In addition, significant differences are present in the tagging process that adds PCR primer sites to genome fragment termini, in preparation of the capturing and blocking DNA fragment pools, and metagenomic GFE applies to a larger range of DNA fragments both in size (150 bp to 1200 bp) and sequence composition (entire genomes and metagenomes). GFE is also substantially different from the currently available SSH genome subtraction method. Unlike SSH, GFE enriches for variable DNA segments using a positive physical selection process. Target DNA segments are isolated by, for example, streptavidin binding and removed from solution, washed, and eluted in a separate reaction. All target DNA strands obtained are then amplified by complementary single-primer PCR (Grothues et al., 1993; NAR 21: 1321-1322). SSH attempts to enrich by an unequal or biased PCR amplification itself, relying on self-complementary terminal regions to suppress amplification of molecules common to both comparison pools. Such PCR-mediated approaches are subject to inherent variability in the PCR process itself, and are not the basis for selecting desired target molecules in GFE.
Gene Fragment Enrichment (GFE) differs from previously known techniques in a variety of ways. SCOTS is a gene expression analysis method, while GFE is intended to determine the differences between microbial genomes and total environmental DNA samples. These approaches are also based on analysis of fundamentally different types of nucleic acids. For example, SCOTS requires the use of difficult RNA extraction methods and reverse transcriptase to make cDNA. This cDNA must then be sorted into bacterial and host nucleic acids by hybridization without competitor other than bacterial rDNA containing plasmid DNA. In GFE, target DNA is first extracted, then sheared by sonication, randomly primed with a Klenow DNA polymerase I reaction, and then amplified by lone-primer-PCR (Grothues et al., 1993; NAR 21: 1321-1322). These are just a few of the differences in these two entirely different procedures.
SCOTS also first requires three initial rounds of selection without blocking competitor in order to obtain the microbial component of cDNA from infected cells or tissues and to normalize the representation toward unit gene copy number. The blocking component of GFE is sheared native microbial DNA, while the blocking cDNA used in the subsequent SCOTS cDNA enrichment are PCR amplicons amplified from a cDNA pool. Unlike SCOTS, GFE has no procedural step or goal to normalize sequences to unit copy number, and there is no need to separate nucleic acids from the host and microbe.
SH is an optimization of Representational Difference Analysis or RDA (Lisitsyn et al. 1993). RDA relies on the difference in amplification efficiency of DNA containing two flanking PCR primer sites (exponential amplification) relative to a single site at one end (linear amplification). By hybridization of DNA strands from two pools of DNA fragments with different linker sequences, those DNA strands from the first pool that are not able to hybridize with strands from the second pool reassociate, and form superior templates for exponential PCR amplification. Hetero-hybrids that form from the annealing of complementary strands from shared DNAs in both pools have only one flanking primer target site, and those that are unique to the second pool do not have any flanking primers sites. Differential amplification of reassociated strands unique to the target pool is then achieved by their exponential increase in a subsequent polymerase chain reaction (PCR). The first is then used to obtain amplified material unique to the first nucleic acid pool. GFE, in contrast, is a physical separation process that relies on competitive hybridization to physically separate nucleic acids prior to PCR amplification (i.e., positive selection process). SH is a PCR mediated selective process, while GFE is a physical separation method followed by an amplification step.
Subtractive hybridization (Straus and Ausubel, 1990; PNAS 87:1889-1893) is a different physical nucleic acid separation process, and relies on the inherently difficult goal of removing all of the common DNA strands from two nucleic acid pools by hybridization (negative selection process). DNA from one source is modified for later selective binding, and is then hybridized with material from a second source. Multiple rounds of hybridization and binding are then used to physically deplete the second pool of all complementary DNA strands. This is a different process from that used in GFE in that it is a negative hybridization and removal process. In contrast, GFE uses a positive selection approach to sample only those nucleic acids that are still able to bind complementary DNA strands in the presence of a competitor from a second source. Unlike these other approaches it does not rely on removing all of the complementary sequences in two nucleic acid pools, as does subtractive hybridization. GFE is therefore inherently less prone to obtaining “false positive” or shared sequences left behind by incomplete subtractive approaches like SSH, RDA, and subtractive hybridization.
The GFE technique thus provides a method for identifying differences between communities of microorganisms. This process includes the following steps:
This process can be used for any microorganisms present in a sample. The sample may originate from any animal suspected of contributing to contamination of a stream or waterway, including but not limited to cattle, fowl, pigs and humans.
The primers can be modified easily using conventional software such as PRIMER EXPRESS from ABI. To modify a primer using this software, one enters the DNA sequence and designates a primer location based on data from the conventional PCR primer. The program then designs a new primer sequence that is modified to work on a real-time platform.
Alternatively, one can modify primer sequences by hand. The key information required is the DNA sequence. It is helpful to have the conventional PCR data to designate where, the 3′ end of the primers should be situated.
Streptavidin is used merely as an illustration of modifying and binding partners that can be used. Any suitable chemical tagging and binding technology will work with GFE.
Genome fragment enrichment is useful in identifying regions of genetic variation between two microbial genomes or metagenomes of entire bacterial communities such as microbiota present in fecal material from different animal species. For microbial genome comparisons, genome fragments from one microbial species are first hybridized with genomic DNA fragments from a second microbial species, and then these fragments are incubated with additional genomic DNA fragments from the first species containing defined sequence tags. The resulting DNA hybrids are captured, and all of the captured strands from the tagged pool are PCR amplified by primers complementary to the added terminal tag sequences. These amplified DNAs are sequences unique to the first microbial species or source. Sequences obtained are then unambiguously identified by cloning into E. coli plasmids and DNA sequencing.
Genome Fragment Enrichment uses a competitive hybridization process that is also a part of the previously described RNA analysis method, SCOTS (Graham et al., 1999). As seen in the second stage of SCOTS, GFE uses competitive solution hybridization to obtain DNA fragments that are present in one pool of fragments but not in another (
Genome fragment enhancement was successfully used to identify hundreds of DNA sequences which are either absent or divergent in one bacterial genome compared to another, as well as microbial cow-specific DNA sequences present in a cow fecal metagenome and absent in a pig metagenome. In addition to cow-specific DNA sequences, GFE has been successfully applied to isolate microbial human-specific and chicken-specific DNA sequences. This technique can also be used to identify DNA sequences either absent or divergent in a variety of bacterial genomes or microbial communities (i.e. fecal samples). Specific non-limiting examples of animals include bovine, human, and avian.
Host-specific primer sets developed from DNA sequences isolated with GFE can be used for end point and real-time PCR applications, as well as microarray applications to make species-specific identifications. Conventional host-specific primer sets can readily be modified to provide real-time PCR primers. Specific non-limiting examples of animals reported herein include cattle, human, and chicken.
Fecal samples were collected from diverse geographic locations in the United States (Florida, New Jersey, West Virginia, Delaware, Ohio, Texas, Nebraska and Georgia) and from one location in the Republic of China (Shandong). Samples from the following animals were used to test the host specificity and host distribution of the potential host-specific markers: Sues scrota (pig), Capra agars (domestic goat), Ovis aries (sheep), Equus caballus (horse). Bos Taurus (bovine), Gallus gallus (chicken), Meleagris gallopavo (turkey), Anser sp. (Canadian goose), Larus californicus (seagull), Treron sp. (pigeon), canis latrans (coyote), Scirus carolinensis (gray squirrel), Odocoileus virginianus (whitetail deer), Didelphis virginiana (possum), Canis familiaris (dog), Felix catus (cat), Lynx rufis (bobcat), Procyon lotor (raccoon), Erinaceus sp. (hedge hog), Loragyps stratus (black vulture) and Homo sapiens (human). These are shown in Table 1.
aDifferent tests were performed to validate the host-specific assays. DNA extracts from fecal samples were used to test host specificity using composite as well as individual fecal samples. To determine host distribution of genetic markers DNA extracts from individual chicken fecal samples were used. DNA extracts from water samples were used to determine the presence of host specific markers in fecally impacted water bodies.
bNumbers in parentheses represent the number of composites used in each particular case. For example, DNA extracts from two sets of five different fecal samples were combined to create the two composites tested in host-specificity studies.
Feces were collected aseptically, placed into sterile conical tubes with screw caps, and stored at −80° C. until required. Total DNA was extracted from the fecal samples using the Mo Bio Fecal kit (Mo Bio Laboratories, CA) or the FastDNA Kit (Q-Biogene, Carlsbad, Calif.) following the protocols provided by the manufacturers. Total DNA was eluted in 50 or 100 microliters of molecular grade water, and DNA concentrations were measured using a Nanoprop® ND-1000 UV-Vis Spectrophotometer (Nanoprop Technologies, Inc., Berlin, Germany, using 2 microliters of each DNA extract.
Water samples that were possibly contaminated by chickens (Georgia), Canadian geese (New Jersey), cow (Nebraska), pig (Ohio) and human (Ohio), shown in Table 1, were collected in sterile bottles and transported to the laboratory in ice coolers. Samples of 100 ml were filtered onto 47 mm polycarbonate membranes (0.2 micron pore size, Millipore corporation, MA). The membranes were then transferred into sterile conical tubes and kept at −80° C. until further processing. DNA from water samples was extracted using the FastDNA Kit.
Genome Fragment Enrichment of Chicken Fecal DNA versus Other Fecal DNA
A modification of the GFE method developed by Shanks et al. was used to enrich for chicken-specific metagenomic regions, shown in
Two independent GFE experiments were performed. In the first experiment, a composite of chicken metagenomic DNA (tester) and a composite of pig metagenomic DNA (blocker) were used to enrich for chicken-specific fragments. This experiment was labeled CP. In order to selectively retrieve fragments from the chicken fecal metagenome, a subsample of the tester DNA was labeled with biotin and used as a capturing surface. To prepare the DNA capturing surface, approximately 10 micrograms of composite fecal DNA from 14 different chickens from West Virginia (714 ng per individual fecal sample) were mechanically sheared into approximately 100-900 base pair (bp) fragments and labeled with biotin. Ten micrograms of a DNA extract composite from nine pigs was used to prepare the blocker solution. To prepare DNA sued to enrich for host-specific fragments, sequence-specific oligonucleotide primers having both a common 5′ sequence and nine random residues (herein called K9 primers; gacactctcgagacatcaccggtacc-nnnnnnnnn) were linked to I microgram of sheared chicken fecal DNA., using Klenow polymerase extension. Metagenomic fragments modified with the K9 primers are herein described as K9-tagged DNA.
The second GFE experiment, labeled CB, was conducted again using chicken metagenomic DNA as the tester, which a composite of metagenomic DNA from different birds (n=35; 11 turkeys, 16 Canadian geese and 8 seagulls) was used as the blocker. The blocker fecal DNA was prepared by mixing equal amounts of DNA for each animal type to reach a total of 10 micrograms of composite DNA.
In both GFE experiments, the pre-hybridization solution was prepared by mixing 100 nanograms of the genomic DNA capturing surface and 1.5 micrograms of the blocker DNA solution overlaid with mineral oil. The solution was heated at 98° C. for two minutes before 4 microliters of 5 M NaCl was added. The mixture was allowed to pre-hybridize for 20 minutes at 55° C. The tagged K9-tagged chicken genomic DNA (1090 ng) was incubated separately at 98° C. for two minutes and mixed with 4 microliters of 5 M NaCl before it was transferred to ice. The entire K9-tagged fecal DNA solution was added to the pre-hybridization solution and was incubated overnight at 55° C. The conditions of the pre-hybridizations and hybridizations in the two experiments for CP and CB were the same. DNA hybrids were isolated by streptavidin binding and the captured K9-tagged genomic fragments were amplified by lone-linker PCR. All PCR reactions were performed using a MJ Research DNA Engine Tetrad 2 thermal cycler (Bio-Rad, Hercules, Calif.). In each case, PCR products from the previous round were used for the next enrichment round. Each experiment was conducted in triplicate using the same preparations of capturing surface, blocking, and tester DNAS. PCR products from five reactions, for each enrichment round, were pooled and cloned into pCR4.1TOPO following the manufacturer's instructions (Invitrogen). Cloning libraries if enriched fragments were developed for rounds 1 and 2 of the CB experiment and for round 2 of the CP experiment.
Individual clones were grown in Luria Broth plus ampicillin as the selected agent, and cells were then added directly to M13-PCR assays to screen for inserts. PCR assays (@5 microliters) contained 1×ExTaq PCR buffer (Panvera), 2.5 mM (each) of dATP, dCTP, dGTP and dTTP; 0.2 microM of M13F and M13R primers; 0.064% bovine serum albumin (Sigma-Aldrich); 0.625 U ExTaq; and 1 microliter of cells. Amplification conditions included an initial incubation at 94° C. for three minutes followed by 20 cycles of 94° C. (30s), 52° C. (20s) and 72° C. (40s). Inserts were confirmed using agarose gel electrophoresis, and PCR products were purified using Qiaquick 96 plate (Qiagen). Sequencing was carried out using Big Dye terminator chemistry and capillary gel electrophoresis (Applied Biosystems PRISM 3730XL DNA Analyzer) at the Cincinnati Children's Hospital Medical Center Genomics Core Facility (Cincinnati, Ohio). Sequences were generated for each clone using M13 forward and reverse primers. Sequence editing and alignment were performed using Sequencer software (Gene Codes Corporation, Ann Arbor, Mich.). The putative protein transcript of each sequence was annotated based on the biochemical function of similar gene sequences using BLASTX with the non-redundant (NR)/GenBank database. BLASTX sequence matches with E values of ≦10−3 and sequence identities of ≧30% were considered to be similar protein sequences. To organize sequences into functional gene categories, the DNA sequences were grouped according to the database of Clusters of Orthologous Groups (COG) of proteins. Enriched sequences were assigned bacterial class annotations based on the top BLASTX hit (lowest E-value score (GenBank NR database.
Primers were designed using Premier Designed software (version 2.01; Cary, N.C.) under the following conditions: no hairpin, no primer dimmer formation, and annealing temperature of 54 or 65° C. Assays were optimized through temperature gradients using various concentrations of fecal DNA templates. Primers were tested for host specificity against fecal DNA composites for each animal type listed above. The primers that showed host specificity to chicken composites were further challenged against fecal DNA extracts from individual cow, human, pig, turkey, Canadian goose, seagull, duck and pigeon specimens (Table 1). Host-specific assays were used to measure host distribution of each genetic marker with individual chicken fecal samples. In addition, selected chicken fecal-specific PCR assays were challenged against DNA extracted from water samples presumed to be impacted with chicken fecal contamination. PCR assays specific to Bacteroidetes spp. and Clostridium coccoides were used to determine the presence of potential PCR inhibitors in all DNA extracts and for the potential presence of fecal pollution in water samples. A positive signal from the general Bacteroides spp. and C. coccoides assays in samples that tested negative for the host-specific assays was used as evidence for the absence of PCR inhibition. All tests were performed using two DNA concentrations including 1 and 10 nanograms/microliter for host-specificity studies, DNA from each fecal sample was first extracted and then equal amounts of each DNA extracts were mixed to create fecal DNA composites. To test host distribution, individual DNA extracts from each targeted animal were tested. The presence of PCR products was visualized using 2% agarose gel electrophoresis and GelStar as the nucleic acid stain (FMC Bioproducts, Rockland, Me.).
A total of 471 clones were characterized in this study, 196 from the CP experiment (chicken fecal DNA versus pig fecal DNA) and 275 from the CB experiment (chicken fecal DNA versus a composite containing turkey, seagull and geese fecal DNA). Eighty sequences did not have significant similarity to sequences in the NR protein database. Based on top BLASTX hits, the analyzed sequences were similar to genes encoded in 19 bacterial groups (classes) and Archae. Clostridia-like sequences were the most abundant group (20.9%), with many sequences showing the greatest similarity (top BLASTX hit based on E-values) to C. perfringens, C. tetani, C. thermocellum and Moorella thermoacetica proteins, as shown in
Bacilli-like sequences were the second most abundant group (17.4%), with sequences similar to members of the genera Enterococcus, Lactobacillus, and Streptococcus. Bacteroides-like sequences represented 15.1% of the total clones analyzed, with sequences showing similarity to Bacteroides spp., Cytophaga hutchinsonii, and Porphyromonas gingivalis. Other sequences showed similarity to proteins from Actinobacteria (8.0%) and Cyanobacteria (4.9) and to α-(1.8%), β-(3.1%), γ-(9.5%) and δ-Proteobacteria (5.8%). A few sequences (1.7%) showed similarity to archael sequences. Interestingly, some of the sequences partially matched pathogenic bacterial genes, such as E. coli (93-100% identity), Listeria monocytogenes (84% identity), Salmonella enterica (96-100% identity), and Shigella flexneri (87-100% identity). Although the sequences in the NR database is biased toward cultural microorganisms, bacterial class designations were made to obtain an idea of the diversity of bacterial populations associated with the enriched DNA libraries.
The abundance of Clostridia- and Bacilli (mainly Lactobacilladales)-like proteins in the metagenomic libraries is not surprising in light of findings from 16S rDNA-based studies indicating that Clostridia and Lactobacilladales represent approximately 65% and 23%, respectively, of the intestinal (cecum) bacterial community of broiler chickens. In contrast, Bacteroidetes are not as numerically dominant as Clostridia and Bacilli in chickens as in other gut systems (e.g., humans), the results obtained further confirm that as a group Bacteroidetes possess a high number of bacterial host-specific genes, some of which might be involved in host-microbial interactions. In addition, these results further confirm the high selection process of the GFE technique.
Other gut bacteria represented in the GFE libraries and commonly identified in 16S rDNA gut libraries are Bifibacterium spp., E. coli, S. Enterica, and Campylobacter spp. Some clones in the GFE library are similar to genes found in environmental bacteria like Arthrobacter spp., Cornebacterium spp., Pseudomonas spp. and Geobacter spp. These organisms might be transitory in the gut and therefore similar sequences would not be good candidate genes for chicken-specific PCR assays.
Most fragments in the top BLASTX sequence match showed significant similarity to a Bacteriodetes protein frequently showed similarity to the same protein in other Bacteroidetes species, suggesting that some of these genes could be playing important roles in this bacterial group. In contrast, in several cases when the sequence matched Clostridia or Bacilli proteins, other potential matches suggested a link to proteins from other organisms, including Paracoccus denitrificans, Cornebacterium efficient, Trichodesmium erythraeumi and Chlamydophila pneumoniae. Altogether, these data suggest that Clostridia and Lactobacillus genes similar to non-fecal bacterial genes might not be good candidates for the development of host-specific PCR assays.
Most fragments in which the top BLASTX sequence match showed significant similarity to a Bacteriodetes protein frequently showed significant similarity to the same protein in other Bacteroidetes species, suggesting that some of these genes could be playing important roles in this bacterial group. In contrast thereto, in several cases when the sequence matched Clostridia or Bacilli proteins, other potential matches suggested a link to proteins from other organisms, including Paracoccus denitrificans, Corynebacterium efficients, Trichodesmium erythraeumi, and Chlamydophilia pneumoniae. Altogether, these data suggest that Clostridia and Lactobacillus genes similar to non-fecal bacterial genes might not be good candidates for the development of host-specific PCR assays.
In both CP and CB libraries, more than 60% of the bacterial clones were similar to poorly characterized genes (e.g., 63.9% were predicted as genes with unknown functions). Sequences similar to clostridia had the largest proportion of enriched fragments associated with uncharacterized functions (56.6%), while Bacteroidetes and Bacilli had a smaller proportion (30.6% and 45.6%, respectively). Of the fragments associated with characterized function genes (36.1% of total analyzed sequences). 55, or 11.0% of the sequences were associated with metabolic process and 63, 13.4%, sequences were associated with information storage and processing (e.g., DNA repair and DNA replication), shown in
Bacteriaroides
fragilis
B. thetaiotoomicron
B. fragilis
Clostridium
thermocellum
Desulfitobacterium
hafniense
B. fragilis
C. tetani
B. fragilis
B. fragilis
C. perfringens str.
B. thetaiotoomicron
B. thetaiotoomicron
B. fragilis
B. fragilis
B. fragilis
B. fragilis
Lactobacillus
acidophilus
Clostridiaceae
B. fragilis
B. fragilis
Bifidobacterium
B. fragilis
C. thermocellum
C. tetani
L. gasseri
Generally, there were no major differences in COG categories between CP and CB libraries, although cellular processes associated with cell motility and metabolic functions associated with lipid transport were present only in the CP library, while metabolic functions associated with nucleotide transport were present only in the CB library, as shown in
Twenty fine sequences were selected from the clone libraries to develop host-specific PCR assays based on the following criteria:
To test host specificity, the PCR assays were challenged against composite fecal DNA extracts form non-target animals and from the chicken composite sample used in the GFE experiments. All primers tested discriminated between testers and blockers (pig in CP and turkey/goose/seagull in CB, respectively). Of the CP-based assays, ten were shown to be chicken specific, three produced amplification products for both chicken and turkey fecal DNA, four assays were positive with fecal DNA from all birds tested (chicken, turkey, seagull and goose), and two cross-reacted with non-avian hosts, as shown in Table 2.
Two assays from the CB sequences were specific to chicken fecal DNA, while the rest also cross-reacted with sheep and/or goat fecal DNA. Of the ten PCR assays targeting cellular related processes, eight of them were shown to be chicken or bird specific. These results are compatible with previous studies showing that functional gene sequences associated with cellular processes are good targets for host-specific PCR assays.
Of the seven DNA sequences possibly related to metabolism, five of these sequences were specific to chicken fecal bacteria (Table 2). Those five DNA sequences were similar to cell membrane proteins involved in the transport of carbohydrates (e.g., CP1-26), inorganic ions (e.g., CP1-24 and CP2-9), amino acids (e.g., CP1-55 and CP3-49), and nucleotides (CP2-10). Sequences related to proteins involved in information storage/processing and pooling characterized functions (CP1-25, CP3-48, CB-R2-42. CP1-40 and CP1-74) were specific to a narrow range of hosts (i.e., chicken and/or turkey). Therefore, these results suggest that host-specific sequences can be found in several COG categories.
Several enriched gene fragments have phylogenic relevance, such as CP1-25, which is similar to the elongation factor G of Clostridium spp. The latter gene is a homolog of elongation factor Tu, a GTP-binding protein that plays a central role in protein synthesis and that has been used as a phylogenetic marker. This is the first phylogenetic gene in addition to the rRNA gene to be potentially useful for developing host-specific markers.
Sequences similar to Bacteroidetes proteins used for PCR assays showed mainly chicken-specific or chicken/turkey-specific signals, while the Clostridia- and Lactobacillus-based PCR assays generated positive signals with fecal DNA of other birds as well. Bacteroidetes 16S rDNA-based methods have been previously shown to discriminate between human and cattle fecal microbial communities; however, 16S rDNA sequences gave not been useful in development of chicken-specific assays.
The present inventors have demonstrated that non-ribosomal Bacteroidetes-like genes were identified as specific to chicken and avian microbial communities using a metagenomic approach. These results are not surprising, as several studies have shown that Bacteroidetes spp. have developed a host-specific relationship with their host. For example, Bacteroidetes spp. are beneficial to the human immune system and human metabolism and can also obtain specific nutritional benefits fro the host gut cells in the form of a diversity of available glycans. Genomic and proteomic data gave provided relevant insights into the nature of the interactions between this commensal bacterial group and the human and mouse gut.
It appears from the studies reported herein that some Bacteroidetes-like populations might also develop host-specific interactions with non-mammalian host types. It also appears that Clostridia- and Bacilli-like proteins might be involved in broad symbiotic interacts, as evidenced by the presence of similar host-specified genetic markers in different avian species.
The geographic distribution of all twelve chicken-specific assays was determined by first challenging each assay against fecal DNA composites of chickens from Delaware, West Virginia, Ohio and Georgia. Differences in the geographic distribution of host-specific markers were observed for most of the PCR assays, as shown in Table 3. For example, all PCR assays produced a positive signal with WV composite chicken fecal samples. It should be noted, however, that WV chicken fecal DNA extracts were used as the tester pool in the GFE experiments. In contrast thereto, several PCR assays did not amplify ‘DNA from fecal samples collected at other sample areas (e.g., CP1-74, CP1-40 and CP3-46). The host distribution of PCR assays that produced positive signals with two or more of the chicken composite fecal DNA templates were challenged against 40 individual chicken feces from the aforementioned geographic locations and 30 individual chicken feces collected in China. Different levels of host distribution were obtained for each of the assays, as shown in Table 3. Three of the PCR assays (CP2-9, CP3-49 and CB-R2-42) amplified at least a third of the individual fecal samples. Nearly half of the chicken samples from China tested positive with three assays, suggesting that these markers are globally distributed and that these PCR assays may be useful for water monitoring in other countries. The PCR assays developed this far were generated by examining a small fraction of the cloned fragments. It is reasonable to expect that the pool of host-specific assays will increase as the number of clones examined increases. The results obtained here indicate that metagenomic enrichments allow for the rapid development of multiple genetic markers to confirm the presence of chicken (or other animal) fecal pollution in waters.
aRefers to specific PCR assay.
bPercentage is the number of positive signals to total fecal samples tested in a certain location.
The detection limit of the CP2-9, CP3-49 and CB-R2-42 assays was determined using fecal DNA extracts from three different locations: West Virginia, Ohio and China. The results show different levels of sensitivity for each of the assays, ranging from 0.001 to 1 ng/microliter of fecal DNA, as shown in Table 4, with the CP2-9-based assay consistently having the highest detection limit. Each assay showed different levels of detection of different fecal samples, suggesting that the density of the population carrying the host-specific markers can vary even among individual hosts, as shown in Table 4. In most cases, the sensitivity of the host-specific assays was lower than assays targeting Bacteroidetes and Clostridia-like sequences (Table 4), suggesting that the markers are found in a subset of the most predominant fecal bacteria, a phenomenon also observed with 16S rDNA-based markers.
Clostridia
a
Bacteroidetes
a
aGeneral PCR assays targeting members of C. coccoides and Bacteroidetes species.
When the three best PCR assays (CP2-9, CP3-49 and CB-R2-42, based on the host distribution results) were challenged against water samples presumed to be contaminated with non-target sources (i.e., cattle, human, pigs, and geese), none of the DNA extracts produced PCR signals, further suggesting the host specificity of these markers. In contrast, when DNA extracts from water samples obtained from Georgia and Delaware watersheds possibly contaminated with chicken feces (see Table 5, Figure *) were used as templates, the CP2-9 assay showed positive signals in one Delaware and two Georgia samples, while the other two assays detected chicken-associated signals in two of the Delaware DNA extracts. None of the markers detected chicken contamination in two Delaware samples. In water samples, the minimal detection limit was 0/1 ng. These results suggest that it may be necessary to use multiple markers when trying to detect any particular source of fecal contamination. Before using any assay in fecal source tracking studies, extensive field testing is required to determine the efficacy of the assays and the geographic distribution of the host-specific markers.
Initially, 70 cow-specific DNA sequences isolated from cow fecal material were identified using the GFE method of the present invention. Three of these sequences were randomly chosen to develop cow discriminatory primer sets, and full scale working applications.
Three randomly selected host-specific Bacteroidales-like GFE sequences were used for host-specific PCR primer development (Table 1). PCR assay 1 was derived from a 368 bp host-specific DNA fragment annotated as a conserved hypothetic secretory protein with an unclassified functional group assignment (locus BT0921). The top BLASTx hit (8.00E−11) for this sequence shared 25% sequence identity to a B. fragils YCH46 hypothetical protein (locus BF2432). Under optimal PCR conditions (62° C. annealing 30 cycles), PCR assay 1 routinely detected fg quantities of cow fecal DNA (
PCR assay 2 targeted a portion of 437 bp fragment annotated as a HDIG domain protein involved in energy metabolism and electron transport (locus BT2749). The top BLASTx hit for PCR assay 2 (32% ID; 1.00E-08) was a B. fragils YCH46 putative membrane-associated HD superfamily hydrolase. Optimal conditions for PCR assay 2 include a 62° C. annealing temperature for 35 cycles, which allowed for the detection of 10 fg cow fecal DNA, as showing
PCR assay 3 originated from a 569 bp fragment encoding for a sialic acid-specific 9-O-acetylesterase secretory protein homologue (locus BT0457) functioning cell envelope biosynthesis and degradation of surface polysaccharides and lipopolysaccharides. The top BLASTx hit for marker 3 (75% ID; 8.00E−80) was a sialate O-acetylesterase protein from B. fraglis YCH46. PCR assay 3 exhibited the lowest limit of detection under optimal conditions (60° C.; 35 cycles) and consistently amplified 0.1 fg of cow fecal DNA (
All host-specific markers amplified template DNA molecules from the original target GFE cow fecal sample, as well as from a large number of individual cow fecal samples not used to construct host-specific GFE clone libraries. Host-specific markers were present in 72% to 91% of 148 cow fecal samples collected from five different geographical locations over a 24-month period (Table 4). PCR assay 3 showed the broadest host distribution and temporal stability by successfully amplifying 91% of all cow fecal samples.
Each primer set was tested against individual non-target DNA molecules. PCR assay 3 exhibited specificity for 99.2% of the fecal samples and only cross-reacted with two alpaca samples. Primer sets demonstrated extremely high levels of specificity in fresh and marine natural water sources. All water samples yielded no PCR product suggesting that indigenous microorganisms from these water sources do not cross-react with host-specified target DNA sequences.
Table 5 provides a summary of host-specific PCR primer sequences, amplicon lengths in base pairs, optimal annealing temperatures (° C.), optimal number of PCR thermal cycles, and limit of detection.
In validation studies, all three cow-specific PCR assays were found to differentiate between cows and 29 other animal species and did not amplify DNA isolated from freshwater and marine microbial communities. These assays also successfully identified cow fecal pollution from water samples collected in two watersheds situated near cow animal feeding operations. Based upon the fact that three randomly chosen sequences worked according to plan, one skilled in the art would expect that the remaining 67 sequences would work just as well. It is also reasonable to expect that human and chicken-specific DNA sequences isolated during GFE will allow for the development of additional human- and chicken-specific PCR assays.
Genome fragment enrichment has been successfully used to identify hundreds of DNA sequences either absent or divergent in one bacterial genome compared to another, as well as microbial cow-, human-, and chicken-specific DNA sequences.
While the protocol described below uses microgram quantities of DNA, GFE has been successfully performed with much smaller starting quantities of DNA. The key to using much smaller quantities of DNA is to maintain specific ratios between target, blocker, and capture surface. It is crucial to use large quantities of blocker DNA relative to the capturing surface for the prehybridization step. In some of the examples in the present specification, approximately 50 times more blocker was used than capturing surface DNA, and one-tenth the amount of capturing surface DNA for target DNA. A lower limit is approximately 1:2 and 1:1 ratios of capture: target. Ideally, one creates a competitive hybridization environment in which the blocked DNA has the advantage, both in quantity of DNA and time, to hybridize to complementary DNA sequences in the capturing surface. This advantage is realized in the prehybridization step, where competitive hybridization of the capturing surface of the capturing surface, the blocking DNA, physically blocks DNA sequences shared between two DNA pools. The unblocked DNA hybridization sites remaining after prehybridization are then available to form DNA hybrids with the terminal tagged target DNA, which is at a disadvantage to the blocker DNA both in quantity of DNA and time to hybridize.
For the comparison of two microbial genomes, E. faecalis genomic DNA 1.8 μg was mechanically sheared by sonication into approximately 150 to 900 base pair (bp) fragments, precipitated in 7.5 M ammonium acetate and 100% ethanol, and dissolved in 15 μg TE (1.0 mM Tris, 0.1 mM EDTA, pH 7.5). DNA was mixed with 1.8 μg of photoactive biotin (PBA; Sigma) and transferred to three 0.2 ml thin wall PCR microtubes in equal volumes to increase the surface area of direct exposure to the light source. Each microtube was placed on ice under a regular 200-watt incandescent light bulb, distance 5 cm, for 20 minutes. The three aliquots were then combined, diluted tenfold with TE (pH 9.0), and extracted with three volumes of n-butanol to remove unincorporated PAB. The supernatant was then discarded, and the remaining solution was split into three equal volumes and concentrated by ammonium acetate and ethanol precipitation.
Blocking DNA can be prepared in any number of ways familiar to one skilled in the art. In the present example, sheared native DNA was used rather than PCR amplified DNA in order to reduce amplification bias in the blocker DNA fragment pool.
To prepare blocking DNA for pre-hybridizing capture fragments as shown in
Four micrograms of E. faecalis genomic DNA were sheared by sonication, precipitated in 7.5 M ammonium acetate and 100% ethanol, and dissolved in 5 μl TE (pH 7.5). Defined terminal sequences were added to these capture target fragments to allow PCR amplification of sequences enriched by competitive hybridization. DNA fragments were re-suspended and incubated at 95° C. for five minutes with 4.5 μg K9-DNA primer (5′GACACTCTCGAGACATCACCGGTACC-NNNNNNNNN-3′). This primer illustrates one of many primers that can be used. The most important characteristics of a primer for use in the present invention are that the sequence works well for T-PCR and to have a random polymer 3′ sequence. The mixture was then cooled on ice for five minutes and primers extended with 50 units DNA polymerase I Klenow fragment as described by the manufacturer (New England BioLabs) for 3.5 hours. Klenow extension products containing tagged termini were purified using a QiaQuick PCR Product Clean-up Kit (Qiagen).
A single primer amplification step was then performed to initially amplify K9-targeted DNA. This has previously been shown to produce a reasonable representation of the original material with DNA fragments of this size. Reactions (100 μl) contained 1×ExTaq PCR buffer (Invitrogen); 2.5 mM each dATP, dCTP, dGTP, and dTTP; 0.2 μM of K9-PCR primer (5′-GACACTCTCGAGACATCACCGG-3′); 1% acetamide; 0.625 U ExTaq, and 10 ng of tagged DNA. Incubation temperatures were 94° C. for 40 seconds, 53° C. for one minute, and 72° C. for 30 seconds, for 28 cycles, followed by a 72° C. extension step lasting 1.5 minutes. PCR products were purified using a QiaQuick PCR Product Clean-up Kit (Qiagen).
A single primer amplification step was then performed to initially amplify K9-tagged target DNA. This has previously been shown to produce a reasonable representation of the original material with DNA fragments of this size (Tarr et al., Journal of Bacteriology 182: 6183-6191, 2000). Reactions of 100 μl each contained 1×ExTaq PCR buffer (Invitrogen), 2.5 mM (each) of dATP, dCTP, dGTP and dTTP, 0.2 μM of K9-PCR primer (5′-GACACTCTCCGAGACATCACCGG-3′), 1% acetamide, 0.625 U Ex Taq, and 10 ng of tagged DNA. As noted above, a different primer can be used, depending upon the terminal sequence used to tag the target DNA. Incubation temperatures were 94° C. for four seconds, 53° C. for one minute, and 72° C. for 30 seconds for 28 cycles followed by a 72° C. extension step for 1.5 minutes. PCR products were purified using a QiaQuick PCR Product Clean-up Kit (Qiagen). All PCR reactions in this study were performed in either low-retention reaction tubes (0.2 ml) or 96-well polypropylene plates using a MJ Research DNA Engine Tetrad 2 thermal cycle.
The temperatures for hybridization used in GFE depend on the physical properties of the DNA used as target, blocker, and capturing surface. Hybridization temperatures from about 40° C. to about 70° C. have successfully been used in GFE.
Two independent full analyses were performed. For each enrichment, 10 μg of blocking E. faecium DNA and 0.6 μg of biotinylated E. faecalis capture DNA were precipitated in ethanol, resuspended in 20 μl EPPS solution (10 mM EPPS, 1 mM EDTA), overlaid with mineral oil, and incubated at 98° C. for two minutes. The incubation temperature was then reduced to 55° C., 4 μl of 5M NaCl were added immediately, and the solution was allowed to self-hybridize for 30 minutes. Five micrograms of K9-tagged E. faecalis PCR product was resuspended in 20 μl of EPPS solution and incubated at 98° C. for two minutes in a second microtube. These two solutions were then mixed together and incubated at 55° C.
Biotinylated DNA hybrids were isolated from the hybridization mixture with Dynabeads M-280 Streptavidin (Dynal Biotech, Brown Deer, Wis.). First, 60 μl of beads were washed with 100 μl B & W buffer (TE, pH 7.5, 2M NaCl) three times. Biotin labeled DBNA was immobilized to the bead surface by mixing washed beads and the hybridization reaction diluted in 500 μl if water at 42° C. for ten minutes. The beads were separated from the diluted hybridization mix with a magnetic particle concentrator (MPC-S; Dynal Biotech (and washed three times with 100 μl SG1 Buffer (0.5 M NaOH, 0.1 M NaCl) and incubated for ten minutes at 37° C. The resulting eluate was then precipitated in ammonium acetate and ethanol and resuspended in 80 μl TE (pH 7.5). Eluted K9-tagged target E. faecalis DNA molecules were selectively amplified as previously described above. The PCR products were purified, pooled, and used as target DNA for a second round of prehybridization and hybridization. The PCR products from the second round were used for a third round.
Initially, it was believed that three rounds of GFE were necessary to isolate unique DNA fragments. However, it has been found that one enrichment round is sufficient.
PCR products from the third round of each independent GFE were incorporated into pCR4.1 TOPO as described by the manufacturer, Invitrogen. Individual clones were then subcultured in 300 μl of Luria Broth containing 10 μg/ml ampicillin, and corresponding plasmid purified prior to screening by PCR for inserts. PCR reactions (25 μl) contained 1×ExTaq PCR buffer (Invitrogen), 2.5 mM (each) dAPT, dCTP, dGTP, and dTTP, 0.2 μM of M13F(5′-GTAAAACGACGGCCAG-3′) and M13R (5′-CAGGAAACAGCTATGCA-3′) primers, 0.064% bovine serum albumin (Sigma), 0.625 U ExTaq and 1 μl of template. Incubation temperatures included 94° C. for three minutes lysis step followed by 20 cycle of 94° C. for 30 seconds, 52° C. for 20 seconds, and 72° C. for 40 seconds. Prior to sequencing, PCR products were purified using Qiaquick 96 Plate (Qiagen). Screening was performed in both directions at the Cincinnati Children's Hospital Medical Center Genomics Core Facility (Cincinnati, Ohio) by the dye-terminator method using an Applied Biosystems PRISM 3730 DNA Analyzer.
To confirm genetic variation in the E. faecalis chromosomal regions identified, dot blot hybridizations were performed with the cloned regions using E. faecium DNA as a probe (Ausubel et al., 2001). PCR products for each enriched DNA sequence were purified using the QiaQuick PCR Purification Kit (Qiagen) and 10 μl of PCR product were denatured with 45 μl of denaturing solution (0.5 M NaOH, 1.5 M NaCl) prior to spotting directly onto nylon membranes (Licor) using a 96-2311 manifold (BioRad). The membranes were neutralized with 10 μl neutralization solution (1M TrisCl pH 8.0, 1.5 M NaCl), and UV cross-linked using a Stratalinker (Stragene) following the manufacturer's instructions. Prehybridization was performed for 1.5 hours at 65° C. in 9 ml of pre-warmed Odyssey DNA Hybridization solution (Licor) containing 1×Denhardt's solution (Sigma) and salmon sperm DNA (Sigma). For probe synthesis, defined terminal sequences were added to E. faecium genomic DNA as described above [GFE (iii) using F9-DNA 5′-GCCGGAGCTCTGCAGAATTC-NNNNNNNNN-3′]. F9-tagged DNA was amplified as described above [GFE (iii) using biotin-16-2′deoxyuridine-5′-triphosphate (Roche) and the F9-PCR primer [5′-GCCGGAGCTCTGCASGAATTC-3′]. The F9-tagged biotin labeled E. faecium PCR product was purified using QiaQuick PCR Purification Kit (Qiagen). Approximately one microgram of probe (20 μl of PCR product) was added to fresh hybridization solution and allowed to hybridize spotted membranes overnight at 55° C. in a rotating hybridization oven. Standard protocols for membrane washing were followed, washing twice under low stringency conditions (room temperature) and twice under moderate stringency conditions (42° C.) (Ausubel et al., 2001). The membranes were visualized with ah Odyssey infrared imaging system (Licor) at an intensity setting of five.
DNA sequence readings were assembled using SeqMan II (DNAstar, Inc.) and compared to the E. faecalis V583 annotated genome at The Institute for Genomic Research (TIGR) with BLASTn. Redundant sequences were removed from the data set. The remaining sequences were then searched against the E. faecium genome draft assembly using the JGI tBLASTx (Joint Genome Institute). The sequences were designated homologous (expectation value ≦1e−03) or absent (no significant hits). Gene attributes were assigned to specific clones based on annotations available at the TIGR comprehensive microbial resource database.
DNA sequence identities between E. faecalis (ATCC #19433) and E. faecalis V583 were calculated using BLASTn (Althschul et al., 1997) generated alignments. Sequence identities between E. faecalis (ATCC #19433) and the E. faecium draft assembly (JGI) were derived from pair wise DNA sequence comparisons using the Wilbur-Lipman method with default settings (MegAlign, DNAstar, Inc.).
Non-redundant clones, false positives, and divergent clones were categorized with cross-species alignments using the JGI BLATn and the E. faecium genome draft assembly database. BLATn was performed with default settings and minimum sequence identity settings of 90% and 80%. Sequences were sorted into two groups using the following criteria:
Summary of E. faecalis GFE Clones
GFE was performed with chromosomal DNA from two enterococcal ATCC type strains. Three hundred total E. faecalis DNA fragments between 163 and 853 base pairs in size were obtained as plasmid inserts following three rounds of GFE in two independent experiments. Analyses of these DNA fragments identified 225 non-redundant sequences (Table 2, GenBank accession numbers CZ191135-CZ191359). Several of these sequences, of 13.7% (n=31) corresponded to variable regions within ribosomal operons, including 16S, 23S and intercistronic spacer regions (ISR) DNA sequences. There are four such operons in the E. faecalis V583 genome (Paulsen et al., 2003). This large number of ribosomal clone sequences may have resulted from PCR kinetics that preferentially amplified the more abundant nucleic acid templates. These non-redundant clones from E. faecalis shared an average of 97.8% sequence with E. faecalis V583, indicating numerous strain-dependent polymorphisms, and only an average of 36% sequence identity with E. faecium (JGI) sequences, as shown in Table 6. The average identify of the enriched clone set to E. faecium was considerably lower than a set of randomly selected E. faecalis V583 genome regions, which showed an average of 58% identity to E. faecium (JGI) draft sequences. Thirty two percent (n=71) of all E. faecalis non-redundant GFE clone sequences were entirely absent from the E. faecium genome draft assembly (JGI).
E. faecalis GFE
E. faecalis
E. faecium
E. faecium
E. faecium
As expected, BLASTn searches against the NCBI GenBank database identified homologous E. faecalis V583 sequences for all 224 non-redundant sequenced clones (Galbraith et al., 2004; Nesbo et al., 2002) (Expectation value cut-off of ≦1×10−6). Only 154 homologous sequences in the E. faecium JGI genome draft assembly could be identified using BLASTp and tBLASTx using an expectation value cut-off of ≦1×10−3). These E. faecalis-specific clone sequences were sorted into nine functional groups based on the annotated complete genome sequence of E. faecalis V583.
The groups consisted of the following:
GFE clone groupings were based on predicted attributes. The percentages of clones conserved across all sequences low-GC Gram-positive bacteria (excluding mycoplasmas, FASTA p-value<10−5) are listed in Table 3. The most frequently assigned gene functional group for all non-redundant GFE clones was the E. Faecalis V583 genome annotated putative surface exposed or membrane associated proteins (22.6%)(Table 7). The percentage of GFE clone sequences conserved across all known low-GC Gram-positive bacteria was only 27.2% for those sequence with an E. faecium homologue, and only 5.6% for clone sequences absent in the E. faecium genome draft assembly (Table 7).
E. faecalis GFE
E. faecium
E. faecium
Thirty four E. faecalis-specific DNA regions were identified by GFE (Table 7) using a more stringent criterion of at least two corresponding GFE clones. For example, five non-redundant GFE clones corresponded to a region predicted to encode for a 5′-nucleotidase family protein and adjacent putative pheromone binding protein (E. faecalis V583, segment 1; region 64,598 to 66,703). Fourteen divergent gene regions potentially encode for proteins annotated as surface exposed or membrane associated open reading frames (Paulsen et al., 2003). In the two independent GFE hybridizations, 76.5% of these 34 DNA regions were identified in both experiments (Table 7), demonstrating good consistency for the method.
Cross-species alignments with JGI BLATn identified 32 false positive final GFE clones (≧90% identity with an E. faecium homologue) and 184 significantly divergent clone regions (≦80% identity with an E. faecium homologue). These 90% and 80% cut-offs were selected based on data from previous genome studies. rDNA clone sequences made up all of these false positives (97%) except for a cell wall surface anchor family protein (locus EFI1896, coordinates 6208-6525). The sequence identity of this cell wall surface anchor family protein was only 89%, but it remained a significant hit with the E. faecium JGI BLATn search because of a 40 bp stretch contained a 90% or greater DNA sequence identity.
Dot blots using E. faecium genomic DNA as a probe identified 62 cross-hybridizing false positive DNA sequences (
Of the 71 clone sequences completely absent in E. faecium (Table 6), only 7 elicited positive results with the dot blot assay (9.8%). These sequences may have very short regions capable of probe hybridization. These results provide experimental confirmation that numerous regions of genetic variation have been identified for these two specific ATCC strains, and are in good agreement with bioinformatic analyses based on the two sequenced strains. Over 90% of GFE sequences absent from the E. faecium genome draft (JGI) also showed no hybridization in the analyses conducted. Dot-blot hybridization also provides a valuable secondary screening method to identify directly GFE false positives when bioinformatic information is not available.
Identification of E. faecalis DNA Sequences Absent or Divergent in E. faecium
Several hundred candidate E. faecalis-specific DNA sequences were obtained by GFE, and specificity was confirmed for a subset of these genomic regions confirmed by dot blot hybridization and a comparative bioinformatic analysis. GFE clones (excluding false positives identified by BLATn) exhibited an average of only 36% DNA sequence identify with. E. faecium sequences (JGI), and approximately one third of these sequences were completely absent in the draft genome. Non-ribosomal GFE sequences also encoded for 34 variable E. faecalis chromosomal regions, of which approximately 75% were independently determined in separate experiments (Table 7).
GFE was found to be a valid approach for identifying genetic differences between closely related microbial genomes.
Using this technique, the following were observed:
While only a fraction of the clones from the enriched GFE libraries were sequenced, it is expected that additional cloning and sequencing would identify additional regions of genetic variation.
GFE Identified Regions of Variation within Highly Conserved DNA Sequences
The most highly conserved and relatively abundant sequences in the E. faecalis V583 genome are four ribosomal RNA (rrn) operons. Thirty one non-redundant clones were isolated, corresponding to ribosomal ISRs (n-3), 23S rRNA genes (n=22), and 16S rRNA genes (n=5). Ribosomal clones shared an average of 95.2% sequence identity with E. faecium homologous sequences, and were classified as false positives using the thresholds described above. However, short regions within 16S, 23S and ISR rrn sequences are also commonly used to differentiate between enterococci species (Patel et al., 1998; Monstein et al., 1998; Williams et al., 1001; Tsiodras et al., 2000; Gurtler et al., 1999; Hall, 1994; Naimi et al., 1997). For example, there are 59 polymorphic nucleotide positions (97.3% sequence identity) between two representative 16S rDNA sequences (Patel et al., 1998). There are only 14 such polymorphic nucleotides in a 300 bp stretch within domain V of the downstream 23S rDNA gene, and all 22 non-redundant 23S r DNA clones obtained by GFE fell within this variable domain V region.
ISR sequences are also widely recognized for their sequence variability and utility in both species identification and strain typing. Previous studies on E. faecalis report 16S-23S ISR sequence citations in the rrn operons, including the presence or absence of a tRNAala gene, and a small number of intraspecies nucleotide substitutions (Gurtler et al., 1999; Hall, 1994; Neimi et al., 1997). Two GFE clones tested contained these previously described 16S-23S ISR sequences, one encoding for the tRNAala gene, and one not. This indicates two distinct E. faecalis rrn operons in the ATCC #19433 strain. No tRNA-containing ISR sequences appeared in the E. faecium draft genome (JGI), as a BLASTn of this ISR sequence was unable to identify any similar sequences. These results demonstrate that GFE was able to obtain previously described species-specific short variable regions within highly conserved DNA sequences.
Genomic regions identified by GFE were predominantly genes predicted to encode for surface associated proteins. E. faecium V583 genome annotations indicating putative surface exposed or membrane associated proteins corresponded to 22.6% of the non-redundant GFE claims (Table 3). The overall frequency of genes annotated as encoding surface associated proteins in the genome is almost three times lower (6.4%). This over-representation suggests that one of the major differences between these species is the composition of proteins associated with the bacterial cell wall. Large sequence variation in genes involved in surface structures has also been observed in Thermotoga maritime (Nesbo et al., 2002) and also among several closely related pathogens (e.g., Escherichia coli 055 and 0157) genomes (Selander et al., 1997; Tarr et al., 2000; Tettelin et al., 2001; Tettelin et al., 2000). These observations are consistent with the idea that the genetic capacity for diverse surface proteins may be characteristic of the differences between closely related microorganisms. Several studies also suggest that this type of variation is due to diversifying selection pressure to avoid different host immune responses (Selander et al., 1997; Tettelin et al., 2000; Maiden et al., 1997). If this class of potential genetic markets does reflect proposed genetic variation, they would be of particular interest for the use of GFE in pathogenicity studies.
Host-Specific PCR Primers Targeting Bacteriodales-Like Sequences Exhibit a Wide Host Distribution Among Cattle
In another study, GFE was used to enrich for DNA fragments isolated from an individual cow fecal metagenome that are absent in a single pig fecal metagenome. Dot blots confirmed the specificity of almost all GFE sequences for the cow fecal metagenome. In addition, three host-specific PCR primer sets were designed and optimized to target randomly selected GFE Bacteriodales-like DNA sequences (Table 8). All primer sets differentiated between the GFE target (cow) and blocker (pig) metagenomic DNA fragment pools, further demonstrating that GFE is a powerful approach for comparing two complex microbial communities. The ability of GFE to isolate DNA sequence between microbial communities was further demonstrated in similar experiments designed to select for microbial DNA sequences specific to human and chicken bacterial fecal communities.
Primers were then tested against 148 cow fecal samples to measure host distribution (Table 7). Host-specific PCR assays routinely amplified more than 80% of the cow fecal samples, regardless of geographic location. Host-specific PCR assays also showed remarkable stability within host animals over a 24-month time period. This surprising geographic and temporal stability, as well as widespread distribution among host populations, was unexpected, considering that GFE was limited to the comparison of two individual fecal samples.
The unexpected stability and broad distribution of the PCR assays in cattle populations led to testing target specificity among other animal groups. All three host-specific primer sets showed extremely high levels of host specificity for cattle (Table 5), and are consistent with 16S rDNA phylogenetic studies reporting the presence of Bacteroidales host-specific endemic subpopulations in various animal fecal samples (Bernard and Field, 2000; Dick, 2005). Data also corroborates the notion that genes encoding for proteins directly involved in host-fecal microbe interactions will exhibit increased levels of specificity over 16S rRNA gene sequences (Simpson, 2002; Scott, 2004; Griffin, 1999; Jiang, 2001). Current 16S rDNA host-specific PCR assays can only discriminate between ruminant and non-ruminant fecal sources (Bernard and Field, 2000). Host-specific primers targeting non-ribosomal sequences differentiated between cattle and five other ruminant or pseudo-ruminant species including goat, sheep, alpaca, llama, and whitetail deer, with the exception of PCR assay 3, which cross-reacted with two alpaca fecal samples.
The present invention provides a widely applicable nucleic acid sorting method and its use in identifying regions of genetic variation between any two preparations of DNA such as two bacterial genomes or samples containing two microbial communities. GFE was able to identify E. faecalis DNA sequences that are absent in E. faecium, as well as cattle-, human-, and chicken-specific DNA sequences that are divergent or absent in other animal species fecal microbial communities. GFE provides a directed alternative to random genome sequencing for identifying genetic variation among bacterial genomes or microbial communities.
It is to be understood that the phraseology or terminology employed herein is for the purpose of description and not of limitation. The means and materials for carrying out various disclosed functions may take a variety of alternative forms without departing from the invention.
Thus, the expressions “means to . . . ” and “means for . . . ” as may be found in the specification above and/or in the claims below, followed by a functional statement, are intended to define and cover whatever structural, physical, chemical, or electrical element or structures which may now or in the future exist for carrying out the recited function, whether or not precisely equivalent to the embodiment or embodiments disclosed in the specification above. It is intended that such expressions be given their broadest interpretation.
The present application is a continuation in part of application Ser. No. 11/316,888, filed Dec. 27, 2005, which claims priority from provisional application Ser. No. 60/686,407, filed Jun. 2, 2005, the entire contents of all of which are hereby incorporated by reference.
| Number | Date | Country | |
|---|---|---|---|
| 60686407 | Jun 2005 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 11316888 | Dec 2005 | US |
| Child | 12429545 | US |
