INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Bovine tuberculosis (bovine TB) caused by Mycobacterium bovis remains a significant disease of farmed cattle in the Republic of Ireland, incurring substantial annual economic costs through surveillance and eradication programs (21, 29). Despite the implementation in this country of a comprehensive testing and eradication scheme operated by the Department of Agriculture and Food since 1954, which has dramatically reduced the incidence of the disease, bovine TB persists within the Irish cattle population (11). Various reasons have been proposed to explain this residual level of infection, including a relatively high level of cattle movement within the country (20) and reservoirs of infection maintained in other domestic, feral, and wild species (15, 21). Although cattle are the definitive host of this pathogen, M. bovis has an exceptionally broad host range and can be maintained in a variety of mammalian species to a greater or lesser extent (35). In Ireland and the United Kingdom, the badger (Meles meles) is known to represent a significant source of infection outside of the cattle population and is thought to be a contributory factor in the repeated herd breakdowns observed in certain areas, although the precise dynamics of TB transmission between badgers and cattle have never been firmly established (17, 20).

Epidemiological analyses of M. bovis infection have been hampered in the past by the difficulty associated with distinguishing between isolates originating from different sources and species. In recent years, molecular biological techniques have been employed to examine M. bovis bacilli at the DNA level and to identify genotypically distinct strain types that may be presumed to be epidemiologically unrelated. Such data may provide information pertinent to routine field investigations undertaken following herd breakdowns and provide a greater understanding of disease transmission within the national cattle herd and between cattle and nonbovine transmission vectors.

Several strategies for typing M. bovis isolates on the basis of DNA polymorphisms have arisen in recent years. Techniques commonly used internationally include restriction fragment length polymorphism (RFLP) analysis (4, 7, 9, 25, 30, 31, 37), spoligotyping (4, 7, 16, 27), pulsed-field gel electrophoresis (12), and PCR-based techniques such as "ampliprinting" (13, 22).

RFLP analysis has been demonstrated to be a robust and highly discriminatory typing procedure due to the availability of multiple DNA probes for the detection of polymorphic loci within the M. bovis genome and has been the method of choice, alongside spoligotyping, for typing M. bovis isolates from cattle and other hosts in Ireland (4, 5, 30, 31). The most commonly used genetic markers for this type of analysis include the mycobacterial insertion sequence IS6110 (33, 34), the highly repetitive polymorphic GC-rich repeat sequence (PGRS) (27), and the direct repeat (DR) region of the mycobacterial genome (14). IS6110 has been widely used as a tool for genotyping Mycobacterium tuberculosis (19, 36), in which it is usually present in up to 20 copies (24). This DNA element has since been adopted for use as an RFLP probe for typing M. bovis isolates also (8, 18). In M. bovis however, IS6110 is generally found in one to five copies, greatly limiting the ability of this element to discriminate between different strains of this organism (12, 30). The majority of Irish bovine-derived M. bovis isolates typed to date contain a single, monomorphic copy of IS6110 which is insufficiently sensitive for epidemiological purposes, necessitating supplementary typing with additional probes (4, 30, 31).

The PGRS-based RFLP probe appears to be the single most discriminatory of the probes currently available for M. bovis strain typing (4, 7, 9, 25) and can be present in up to 30 copies in members of the M. tuberculosis complex (23, 28). PGRS is present in multiple copies interspersed throughout the genome and exhibits a high level of polymorphism between unrelated isolates. However, the result of a PGRS DNA fingerprint is relatively complex, containing a great many bands that can be difficult to interpret and which can hinder computer-assisted band analysis, particularly when the final image is less than ideal.

The third most commonly used RFLP probe for this kind of analysis is the DR region. Unique to members of the TB complex, the DR region consists of a series of identical 36-bp DR sequences interspersed with variable spacer sequences from 35 to 41 bp in length (14). This region has found application both as a target for RFLP typing and for the PCR amplification-based spoligotyping technique, both of which give a satisfactory level of strain discrimination (1, 7). The DR RFLP probe is targeted at the 36-bp DR sequences found in multiple copies within the DR region, whereas spoligotyping targets the spacer sequences in between. As both DR RFLP analysis and spoligotyping target the same genomic locus, albeit different areas within that same locus, the results of these two methods tend to be comparable in terms of strain discrimination and sensitivity (8). While having the advantages of being a considerably faster and less labor-intensive method to perform than RFLP analysis, spoligotyping alone does not usually provide sufficient discrimination between strains of M. bovis to be used as a sole typing method and is often combined with supplementary techniques such as RFLP typing with IS6110 and PGRS (4, 8, 26).

A recent geographical survey of Irish M. bovis isolates in which 452 M. bovis isolates were examined by RFLP analysis with IS6110, PGRS, and DR and also by spoligotyping (4) showed that 20% of isolates analyzed exhibited a common RFLP strain type, designated strain type A1 A1 A (IS6110, PGRS, and DR, respectively). This strain type, which cannot be further subdivided by existing RFLP methods, occurred in a variety of host species and was widely distributed across the country, appearing in geographically distant regions of the country where genetically isolated genotypes might be expected to arise. This strain type is similar to the most frequently occurring strain type identified from isolates in Northern Ireland (30). This may be due to an overall genetic homogeneity within the Irish M. bovis population and may represent a significant limitation on the usefulness of strain typing as an aid to the epidemiological investigation of herd breakdowns. As the collection and cultivation of isolates for typing is time-consuming and labor-intensive, there has been a need for supplementary methods of dividing this common type in order to maximize the amount of data that can be obtained from cultured material.

In this study, a new RFLP probe has been identified from an M. bovis genomic DNA library which is capable of splitting the most common M. bovis strain type observed in Irish isolates. Preliminary analysis indicates that this probe yields a clear, highly polymorphic banding pattern that has a discriminatory ability comparable to the current combination of IS6110, PGRS, and DR probes.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

The definitions suggested by Tenover et al. (32) have been adopted throughout this study. The term "isolate" refers to cultured bacteria grown from a single colony taken from a primary isolation plate, and an isolate is presumed to be derived from a single organism. Epidemiologically related isolates are isolates collected from a well-defined area as part of an epidemiological investigation that suggests that they may be derived from a common source. Similarly, "strain type" refers to an isolate or group of isolates that can be distinguished from other isolates by genotypic characteristics and may be considered a descriptive subdivision of a species.

Generation of an M. bovis genomic DNA library. Total genomic DNA was extracted from cultured M. bovis bacilli as described previously (31). Genomic DNA was partially digested with the restriction enzyme MboI and fragments in the 5-to-6-kb size range were recovered by preparative gel electrophoresis (2). Purified genomic fragments were ligated into BamHI-digested plasmid pGEM4Z (Promega) and transformed into Escherichia coli DH10B cells (Gibco BRL). Transformants were plated onto 24- by 24-cm agar plates containing ampicillin at 100 µg/ml overlaid with a positively charged nylon membrane (Roche Molecular Biochemicals) and grown overnight at 37°C.

Primary library screening. Approximately 3,000 recombinant E. coli colonies were screened for potentially repetitive mycobacterial genomic inserts. Colonies were lysed in situ, and plasmid DNA was covalently bound to nylon by UV cross-linkage. Filters were hybridized with total M. bovis genomic DNA labeled with digoxygenin (DIG; Roche) at a concentration of 30 ng per ml of hybridization buffer. Due to the highly repetitive nature of PGRS, steps were taken to minimize the changes of detecting recombinant clones containing PGRS by masking these sequences through the inclusion of unlabeled PGRS (in the form of plasmid pTBN12 [28]) at 100 µg/ml in both prehybridization and hybridization solutions. Prehybridization and hybridization solutions also contained unlabeled E. coli chromosomal DNA at 100 µg/ml to prevent cross-hybridization of M. bovis and E. coli genomic DNA. Prehybridization of membranes with unlabeled DNA was allowed to proceed for 4 h. Hybridization with labeled M. bovis genomic DNA was limited to 4 h to allow the preferential binding of repetitive elements over unique sequence, and signals were recorded by a brief flash exposure to X-ray film to detect the most strongly hybridizing colonies. All probes were labeled using the nonradioactive DIG system (Roche), and signal detection was carried out using anti-DIG alkaline phosphatase and the chemiluminescent substrate CDP-Star according to the manufacturer's instructions (Roche).

Secondary library screening. Recombinant colonies that generated hybridization signals with M. bovis genomic DNA above background levels in duplicate screens were subcultured and subjected to a secondary screening process. Mycobacterial genomic inserts were excised by EcoRI and HindIII digestion and probed sequentially with M. bovis genomic DNA and the PGRS sequence contained within plasmid pTBN12. Inserts that gave strong hybridization signals with genomic DNA without binding PGRS were selected for further analysis. Clones that generated a hybridization signal with PGRS were discarded.

Probe generation. Both the PGRS and DR probes were obtained as commercially synthesized oligonucleotides based on the consensus sequence of the PGRS repetitive element (5' CCG TTG CCG CCG TTG CCG CCG TTG CCG CCG) (10) and the DR 36-bp repeat (5' GTC GTC AGA CCC AAA ACC CCG AGA GGG GAC GGA AAC) (14), respectively. A DIG label was incorporated into these probes during synthesis. IS6110 was amplified by PCR using the primer 5' ACG TAC GAA TTC TGA ACC GCC CCG G 3' on the basis of its terminal inverted repeat sequence (34). The 1.3-kb IS6110 region was labeled with DIG by random primed DIG labeling (Hi-Prime DIG labeling Kit; Roche) and used as a probe in its entirety. Plasmid pUCD was extracted and purified by standard methods, and the complete plasmid was labeled with DIG, as the pGEM4Z vector backbone of pUCD had previously been hybridized to M. bovis DNA and did not generate a hybridization pattern.

RFLP methods. Mycobacterial DNA extraction was carried out using cetyltrimethylammonium bromide extraction protocol as described previously (31). Genomic DNA (1 µg) was digested overnight at 37°C with 10 U of the restriction enzyme AluI for strain typing with pUCD, PGRS, and DR and with 10 U of PvuII for strain typing with IS6110. Digested DNA was electrophoresed for 12 to 16 h in 25-cm 1% agarose gels at 2 V/cm in Tris-borate-EDTA (TBE) electrophoresis buffer, transferred to a positively charged nylon membrane (Roche) by overnight Southern capillary blotting, and covalently bound by UV cross-linkage. Internal molecular size ladders were not included; instead, a DIG-labeled molecular size ladder (Roche Molecular Size Ladder VII) was included in the outside lanes. Hybridization was carried out overnight in a 50% formamide buffer at 42°C.

M. bovis isolates. Nylon membranes containing AluI-digested genomic DNA samples from M. bovis isolates originating from different geographical regions of the country and from different host species were available for analysis with pUCD following routine typing. Isolates were obtained during the period 1993 to 1998 as part of routine epidemiological investigations and were identified as M. bovis by standard biochemical tests as described previously (6). Isolates were subcultured on 50 ml of Middlebrook 7H9 medium with oleic acid, albumin, dextrose, citric acid enrichment (Difco, Detroit, Mich.) for 4 weeks at 37°C and were harvested after inactivation by heating at 75°C for 1 h.

Nomenclature. The RFLP naming system adopted by Costello et al. (4) was used throughout this study, where each strain type is assigned an alphanumeric designation. Spoligotype patterns were identified by a letter and a number, and pUCD strain types were identified numerically.

Characterization of plasmid clone pUCD. The M. bovis genomic insert contained within plasmid pUCD was partially characterized by subcloning and sequencing. The insert DNA was digested with AvaI to produce fragments of suitable size for subcloning. AvaI fragments were blunted by incubation with Taq DNA polymerase in the presence of deoxynucleoside triphosphates for 15 min at 70°C, ligated into a T vector (pGEM T-Easy; Promega), and transformed into E. coli DH10B cells (Gibco BRL). Recombinants were subcultured, and the plasmid DNA was extracted and purified for sequencing. DNA sequencing was carried out using an automated DNA sequencer (model 370A; Applied Biosystems).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Recovery of repetitive DNA elements from a genomic library. A genomic DNA plasmid library prepared from M. bovis was screened for repetitive sequence elements by selecting recombinant clones that generated a strong hybridization signal with total M. bovis genomic DNA after a brief hybridization. Twenty recombinant E. coli colonies that exhibited hybridization signals above background levels were selected in this manner, 14 of which were examined for potential value as RFLP probes.

Differentiation of the A1 A1 A strain type. Plasmid pUCD subdivided 21 isolates of the prevalent strain type A1 A1 A into seven different banding patterns, as illustrated in Fig. 1A. These isolates were confirmed as being type A1 A1 A by typing with IS6110, PGRS, and DR, and the banding patterns generated by PGRS are reproduced in Fig. 1B for comparison. Known epidemiological links between the isolates were conserved among the strain types produced by pUCD, as demonstrated in Table 1. It can be seen that A1 A1 A isolates of different geographical origin have been successfully split with pUCD but remain grouped with IS6110, PGRS, and DR and also by spoligotyping, whereas isolates that originated from a single property remain identical.

View larger version (98K): [in this window] [in a new window]   FIG. 1.   (A) Twenty-one M. bovis isolates of the predominant strain type A1 A1 A subdivided by pUCD into seven strain types. pUCD strain types are shown underneath the panel. Arrows indicate some of the background banding evident with this probe that would not be included in strain type identification. Data relating to these isolates are listed in Table 1. MW, molecular weight. (B) Banding pattern produced when the same isolates are probed with PGRS. Molecular sizes are shown to the left. Images were acquired using a Microtek flatbed scanner and labeled using Adobe Photoshop version 4.0.

Performance comparison of pUCD with IS6110, PGRS, and DR RFLP analysis and spoligotyping. Sixty epidemiologically related and nonrelated M. bovis isolates from across the country were typed by spoligotyping and RFLP analysis using IS6110, PGRS, DR, and pUCD (Fig. 2). Spoligotyping of these isolates resulted in 8 strain types. RFLP analysis using IS6110, PGRS, and DR identified 10, 11, and 10 strain types, respectively, and in combination split the group into 18 different strain types. pUCD typing alone resulted in 19 strain types. This group included an additional 13 isolates of the A1 A1 A strain type, which were subdivided into six strain types with pUCD.

View larger version (112K): [in this window] [in a new window]   FIG. 2.   Patterns generated with 60 M. bovis isolates by RFLP analysis using pUCD. The strain type identification is listed underneath each panel. Lanes 31 and 62 contain M. tuberculosis control strain 14323. Details relating to these isolates are listed in Table 1. Molecular sizes are shown to the left. Images were acquired using a Microtek flatbed scanner and labeled using Adobe Photoshop version 4.0.

Characterization of pUCD. The genomic insert contained within plasmid pUCD was subcloned and partially sequenced in order to characterize the DNA element(s) responsible for generating the polymorphism observed between M. bovis isolates. The sequence data obtained from three subclones were compared with the M. tuberculosis genome database maintained at the Sanger Centre (http://www.sanger.ac.uk/Projects/M_tuberculosis) using the BLAST search algorithm. Each subclone was aligned within a 4-kb region of the M. tuberculosis genome (bases 2163227 to 2166681) with >90% identity, indicating that a homologous region of the M. bovis genome is likely represented within pUCD. Subsequent bioinformatics analysis and comparison of the sequence data against the M. tuberculosis complete genome database at GenBank identified segment 85/162 of the M. tuberculosis H37Rv complete genome (3) as an analogous region (accession no. Z97193, bases 40021 to 43561). This region was identified as part of the PPE gene family, found widely dispersed throughout the mycobacterial genome (3). This sequence includes a series of five near-perfect 69-bp repeat units with the following sequence: TCGGTCCGATTGTGGTGCCGGATAT TACT T TCCTGGTAT TCCGTTGAGCCTGAACGCGCTGGGTGGTG. This region also contained three copies of a second repeat unit with the following sequence: TTAATCCGTTCGGTTTGAATATTCCGCTGAGCGGTGCTACCAACGCTGTCACGATCCCTGGTTTCGCGA. A subsequent BLAST search of this region against the M. tuberculosis database produced significant alignments with sequences also identified as PPE gene partial coding sequences (accession no. AF082288, AF082289, and AF086633) which contained similar repeat units.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Bovine TB remains a significant economic burden in Ireland despite 40 years of intensive control measures aimed at ensuring its complete eradication from the cattle population. Various authorities have concluded that, in countries where there is transmission of infection from spillover hosts into cattle, complete eradication is not a feasible goal and control measures must be applied indefinitely (11, 21, 29). However, the pathogen has been successfully eliminated in other European countries through similar practices, and the goals of the Department of Agriculture's eradication program remain unchanged. A more complete understanding of the epidemiology of M. bovis and its various routes of transmission into cattle will complement ongoing eradication strategies. The application of molecular typing technology is currently the most effective means of investigating this.

In a recent proposal for a standardized approach to DNA fingerprinting of M. bovis from an unknown population, Cousins et al. (8) recommended that isolates should be typed in the first instance with IS6110 in order to define the population, followed by either spoligotyping or RFLP analysis with PGRS, DR, or a combination of both when greatest sensitivity is required. The study recommended that, when isolates contain three or more copies of the IS6110 element, IS6110 RFLP typing should be sufficient for epidemiological purposes. In countries where the majority of bovine M. bovis isolates generally contain a single copy of IS6110, such as Ireland (4), Argentina (25), and Australia (7), genotyping with the IS6110 probe alone does not generate sufficient polymorphism to be of value to epidemiological studies, and secondary probes such as PGRS and DR are necessary to obtain suitable levels of discrimination. Whereas IS6110 generates maximal polymorphism when used to probe PvuII-digested genomic DNA, the restriction enzyme of choice for PGRS and DR analysis is AluI. The need for a second restriction enzyme when working with PGRS and DR necessitates the preparation of a second set of samples. In addition, samples to be analyzed with PGRS and DR should ideally be prepared separately, as each probe requires different electrophoresis conditions for optimum separation of the resulting bands. With PGRS the bands of interest are relatively large (2 to 5 kb), whereas DR detects much smaller bands (300 to 1,100 bp). This makes RFLP analysis a labor-intensive technique in situations where strain discrimination of the highest resolution is required, such as in areas that present a genetically homogeneous M. bovis population.

By screening a genomic library for potentially repetitive sequences, we have identified a region of the M. bovis chromosome which, used as an RFLP probe, may prove to be a useful adjunct to current genotyping protocols. This region, when hybridized to AluI-digested genomic DNA, generates a clear, highly polymorphic banding pattern which effectively split the common A1 A1 A strain type observed within Irish M. bovis isolates. A total of 34 A1 A1 A isolates were examined in this study, and these were subdivided into 10 strain types by pUCD. In addition, when used to strain type 60 isolates, pUCD detected a greater number of strain types than did IS6110, PGRS, DR, or spoligotyping and one more strain type than IS6110, PGRS, and DR in combination. As genomic DNA to be analyzed with both pUCD and DR is prepared using the restriction enzyme AluI, there is the option of using both of these probes consecutively on the same blot, as the gel running conditions normally applied for DR typing are also suitable for pUCD in terms of band spacing and definition. A pUCD-DR combination increased the total number of strain types detected within this study to 26. The restriction enzyme AluI was chosen for preparing template DNA, as membranes containing AluI-digested isolates were readily available from previous routine typing studies. pUCD has not as yet been assessed with different restriction enzymes which could potentially increase the discriminatory power of the probe, although such a change would preclude the subsequent use of the DR probe on the same membrane.

Discriminatory power alone is only one measure of the usefulness of an RFLP probe for epidemiological purposes and must be backed up by concordance with known data concerning specific disease outbreaks. The results observed for the set of isolates used in this study conformed well with known epidemiological information relating to the samples. A proportion of herd breakdowns will generally exhibit multiple strain types, particularly when several probes are used in the analysis, which may reflect infections with more than one clonal orgin. An example of this can be seen in Table 1, where samples 39, 40, and 41 represent three isolates originating from a single herd breakdown from County Limerick, Ireland. Isolates 39 and 40 were classed as pUCD strain type 20, whereas isolate 41 was identified as strain type 11, which suggests an additional route of infection into this herd. This occurrence was reiterated in samples 25, 26, 27, and 28, which originated from a single property in County Waterford, Ireland. (Table 1). In this case the DR probe identified a different strain type for sample 25.

The potential value of pUCD is illustrated in cases where M. bovis isolates cultured from neighboring herds or herd breakdowns in which an epidemiological link is not suspected are identical when genotyped with existing probes but may be distinguished by pUCD. An example of such a situation is contained within the sample of M. bovis isolates used in this study where isolates 19, 20, and 21 (Table 1; Fig. 1A) were obtained from herds in the same local region (County Westmeath, Ireland). Isolates 19 and 20 originated from the same herd, and isolate 21 originated from a neighboring property in the same local area. All three isolates were identical by RFLP analysis using IS6110, PGRS, and DR and also by spoligotyping, whereas RFLP analysis using pUCD displayed a different banding pattern for isolate 21.

The presence of a region homologous to the M. bovis-derived pUCD clone within the M. tuberculosis genome database suggests that this probe may also be of use for strain typing other mycobacterial species. M. tuberculosis 14323 was used as a control strain within this study and produced a banding pattern with pUCD, as would be expected. The polymorphic banding pattern produced by pUCD is suggestive of one or more repeated sequences approximately 50 to 100 bp long. The presence of ~69-bp repeat units within the DNA sequences identified from the M. tuberculosis genome database seems to support this argument. In addition to the polymorphic bands of interest, the pattern displayed by pUCD includes a proportion of bands which do not contribute to strain differentiation. These bands are presumably derived from additional sequence elements contained within the cloned region, and it is hoped that further characterization of pUCD will allow the elimination of this background banding, simplifying the resulting pattern further.

The results reported here suggest that the RFLP probe pUCD provides discriminatory power equivalent to that of a IS6110-PGRS-DR probe combination in a single hybridization step, with a concomitant reduction in the time taken to obtain a final result. The discriminatory power of this probe along with its ability to subdivide the A1 A1 A strain type suggests that it will prove to be a useful epidemiological tool in the study of bovine TB transmission in cases where current probes reveal a high proportion of a common strain type.